From e620c527860a004e1d5ef160b68854f201002af8 Mon Sep 17 00:00:00 2001
From: ModelHub XC <noreply@modelhub.org.cn>
Date: Fri, 3 Jul 2026 05:54:17 +0800
Subject: [PATCH] =?UTF-8?q?=E5=88=9D=E5=A7=8B=E5=8C=96=E9=A1=B9=E7=9B=AE?=
 =?UTF-8?q?=EF=BC=8C=E7=94=B1ModelHub=20XC=E7=A4=BE=E5=8C=BA=E6=8F=90?=
 =?UTF-8?q?=E4=BE=9B=E6=A8=A1=E5=9E=8B?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Model: roonbug/jj75i299
Source: Original Platform
---
 .gitattributes                                |    56 +
 README.md                                     |   201 +
 chat_template.jinja                           |     4 +
 checkpoint-1000/chat_template.jinja           |     4 +
 checkpoint-1000/config.json                   |    77 +
 checkpoint-1000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-1000/model.safetensors.index.json  |   472 +
 checkpoint-1000/optimizer.pt                  |     3 +
 checkpoint-1000/rng_state.pth                 |     3 +
 checkpoint-1000/scheduler.pt                  |     3 +
 checkpoint-1000/special_tokens_map.json       |    34 +
 checkpoint-1000/tokenizer.json                |     3 +
 checkpoint-1000/tokenizer.model               |     3 +
 checkpoint-1000/tokenizer_config.json         |  2013 +++
 checkpoint-1000/trainer_state.json            |  1254 ++
 checkpoint-1000/training_args.bin             |     3 +
 checkpoint-10000/chat_template.jinja          |     4 +
 checkpoint-10000/config.json                  |    77 +
 checkpoint-10000/generation_config.json       |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-10000/model.safetensors.index.json |   472 +
 checkpoint-10000/optimizer.pt                 |     3 +
 checkpoint-10000/rng_state.pth                |     3 +
 checkpoint-10000/scheduler.pt                 |     3 +
 checkpoint-10000/special_tokens_map.json      |    34 +
 checkpoint-10000/tokenizer.json               |     3 +
 checkpoint-10000/tokenizer.model              |     3 +
 checkpoint-10000/tokenizer_config.json        |  2013 +++
 checkpoint-10000/trainer_state.json           | 12234 ++++++++++++++++
 checkpoint-10000/training_args.bin            |     3 +
 checkpoint-1500/chat_template.jinja           |     4 +
 checkpoint-1500/config.json                   |    77 +
 checkpoint-1500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-1500/model.safetensors.index.json  |   472 +
 checkpoint-1500/optimizer.pt                  |     3 +
 checkpoint-1500/rng_state.pth                 |     3 +
 checkpoint-1500/scheduler.pt                  |     3 +
 checkpoint-1500/special_tokens_map.json       |    34 +
 checkpoint-1500/tokenizer.json                |     3 +
 checkpoint-1500/tokenizer.model               |     3 +
 checkpoint-1500/tokenizer_config.json         |  2013 +++
 checkpoint-1500/trainer_state.json            |  1864 +++
 checkpoint-1500/training_args.bin             |     3 +
 checkpoint-2000/chat_template.jinja           |     4 +
 checkpoint-2000/config.json                   |    77 +
 checkpoint-2000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-2000/model.safetensors.index.json  |   472 +
 checkpoint-2000/optimizer.pt                  |     3 +
 checkpoint-2000/rng_state.pth                 |     3 +
 checkpoint-2000/scheduler.pt                  |     3 +
 checkpoint-2000/special_tokens_map.json       |    34 +
 checkpoint-2000/tokenizer.json                |     3 +
 checkpoint-2000/tokenizer.model               |     3 +
 checkpoint-2000/tokenizer_config.json         |  2013 +++
 checkpoint-2000/trainer_state.json            |  2474 ++++
 checkpoint-2000/training_args.bin             |     3 +
 checkpoint-2500/chat_template.jinja           |     4 +
 checkpoint-2500/config.json                   |    77 +
 checkpoint-2500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-2500/model.safetensors.index.json  |   472 +
 checkpoint-2500/optimizer.pt                  |     3 +
 checkpoint-2500/rng_state.pth                 |     3 +
 checkpoint-2500/scheduler.pt                  |     3 +
 checkpoint-2500/special_tokens_map.json       |    34 +
 checkpoint-2500/tokenizer.json                |     3 +
 checkpoint-2500/tokenizer.model               |     3 +
 checkpoint-2500/tokenizer_config.json         |  2013 +++
 checkpoint-2500/trainer_state.json            |  3084 ++++
 checkpoint-2500/training_args.bin             |     3 +
 checkpoint-3000/chat_template.jinja           |     4 +
 checkpoint-3000/config.json                   |    77 +
 checkpoint-3000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-3000/model.safetensors.index.json  |   472 +
 checkpoint-3000/optimizer.pt                  |     3 +
 checkpoint-3000/rng_state.pth                 |     3 +
 checkpoint-3000/scheduler.pt                  |     3 +
 checkpoint-3000/special_tokens_map.json       |    34 +
 checkpoint-3000/tokenizer.json                |     3 +
 checkpoint-3000/tokenizer.model               |     3 +
 checkpoint-3000/tokenizer_config.json         |  2013 +++
 checkpoint-3000/trainer_state.json            |  3694 +++++
 checkpoint-3000/training_args.bin             |     3 +
 checkpoint-3500/chat_template.jinja           |     4 +
 checkpoint-3500/config.json                   |    77 +
 checkpoint-3500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-3500/model.safetensors.index.json  |   472 +
 checkpoint-3500/optimizer.pt                  |     3 +
 checkpoint-3500/rng_state.pth                 |     3 +
 checkpoint-3500/scheduler.pt                  |     3 +
 checkpoint-3500/special_tokens_map.json       |    34 +
 checkpoint-3500/tokenizer.json                |     3 +
 checkpoint-3500/tokenizer.model               |     3 +
 checkpoint-3500/tokenizer_config.json         |  2013 +++
 checkpoint-3500/trainer_state.json            |  4304 ++++++
 checkpoint-3500/training_args.bin             |     3 +
 checkpoint-4000/chat_template.jinja           |     4 +
 checkpoint-4000/config.json                   |    77 +
 checkpoint-4000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-4000/model.safetensors.index.json  |   472 +
 checkpoint-4000/optimizer.pt                  |     3 +
 checkpoint-4000/rng_state.pth                 |     3 +
 checkpoint-4000/scheduler.pt                  |     3 +
 checkpoint-4000/special_tokens_map.json       |    34 +
 checkpoint-4000/tokenizer.json                |     3 +
 checkpoint-4000/tokenizer.model               |     3 +
 checkpoint-4000/tokenizer_config.json         |  2013 +++
 checkpoint-4000/trainer_state.json            |  4914 +++++++
 checkpoint-4000/training_args.bin             |     3 +
 checkpoint-4500/chat_template.jinja           |     4 +
 checkpoint-4500/config.json                   |    77 +
 checkpoint-4500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-4500/model.safetensors.index.json  |   472 +
 checkpoint-4500/optimizer.pt                  |     3 +
 checkpoint-4500/rng_state.pth                 |     3 +
 checkpoint-4500/scheduler.pt                  |     3 +
 checkpoint-4500/special_tokens_map.json       |    34 +
 checkpoint-4500/tokenizer.json                |     3 +
 checkpoint-4500/tokenizer.model               |     3 +
 checkpoint-4500/tokenizer_config.json         |  2013 +++
 checkpoint-4500/trainer_state.json            |  5524 +++++++
 checkpoint-4500/training_args.bin             |     3 +
 checkpoint-500/chat_template.jinja            |     4 +
 checkpoint-500/config.json                    |    77 +
 checkpoint-500/generation_config.json         |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-500/model.safetensors.index.json   |   472 +
 checkpoint-500/optimizer.pt                   |     3 +
 checkpoint-500/rng_state.pth                  |     3 +
 checkpoint-500/scheduler.pt                   |     3 +
 checkpoint-500/special_tokens_map.json        |    34 +
 checkpoint-500/tokenizer.json                 |     3 +
 checkpoint-500/tokenizer.model                |     3 +
 checkpoint-500/tokenizer_config.json          |  2013 +++
 checkpoint-500/trainer_state.json             |   644 +
 checkpoint-500/training_args.bin              |     3 +
 checkpoint-5000/chat_template.jinja           |     4 +
 checkpoint-5000/config.json                   |    77 +
 checkpoint-5000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-5000/model.safetensors.index.json  |   472 +
 checkpoint-5000/optimizer.pt                  |     3 +
 checkpoint-5000/rng_state.pth                 |     3 +
 checkpoint-5000/scheduler.pt                  |     3 +
 checkpoint-5000/special_tokens_map.json       |    34 +
 checkpoint-5000/tokenizer.json                |     3 +
 checkpoint-5000/tokenizer.model               |     3 +
 checkpoint-5000/tokenizer_config.json         |  2013 +++
 checkpoint-5000/trainer_state.json            |  6134 ++++++++
 checkpoint-5000/training_args.bin             |     3 +
 checkpoint-5500/chat_template.jinja           |     4 +
 checkpoint-5500/config.json                   |    77 +
 checkpoint-5500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-5500/model.safetensors.index.json  |   472 +
 checkpoint-5500/optimizer.pt                  |     3 +
 checkpoint-5500/rng_state.pth                 |     3 +
 checkpoint-5500/scheduler.pt                  |     3 +
 checkpoint-5500/special_tokens_map.json       |    34 +
 checkpoint-5500/tokenizer.json                |     3 +
 checkpoint-5500/tokenizer.model               |     3 +
 checkpoint-5500/tokenizer_config.json         |  2013 +++
 checkpoint-5500/trainer_state.json            |  6744 +++++++++
 checkpoint-5500/training_args.bin             |     3 +
 checkpoint-6000/chat_template.jinja           |     4 +
 checkpoint-6000/config.json                   |    77 +
 checkpoint-6000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-6000/model.safetensors.index.json  |   472 +
 checkpoint-6000/optimizer.pt                  |     3 +
 checkpoint-6000/rng_state.pth                 |     3 +
 checkpoint-6000/scheduler.pt                  |     3 +
 checkpoint-6000/special_tokens_map.json       |    34 +
 checkpoint-6000/tokenizer.json                |     3 +
 checkpoint-6000/tokenizer.model               |     3 +
 checkpoint-6000/tokenizer_config.json         |  2013 +++
 checkpoint-6000/trainer_state.json            |  7354 ++++++++++
 checkpoint-6000/training_args.bin             |     3 +
 checkpoint-6500/chat_template.jinja           |     4 +
 checkpoint-6500/config.json                   |    77 +
 checkpoint-6500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-6500/model.safetensors.index.json  |   472 +
 checkpoint-6500/optimizer.pt                  |     3 +
 checkpoint-6500/rng_state.pth                 |     3 +
 checkpoint-6500/scheduler.pt                  |     3 +
 checkpoint-6500/special_tokens_map.json       |    34 +
 checkpoint-6500/tokenizer.json                |     3 +
 checkpoint-6500/tokenizer.model               |     3 +
 checkpoint-6500/tokenizer_config.json         |  2013 +++
 checkpoint-6500/trainer_state.json            |  7964 ++++++++++
 checkpoint-6500/training_args.bin             |     3 +
 checkpoint-7000/chat_template.jinja           |     4 +
 checkpoint-7000/config.json                   |    77 +
 checkpoint-7000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-7000/model.safetensors.index.json  |   472 +
 checkpoint-7000/optimizer.pt                  |     3 +
 checkpoint-7000/rng_state.pth                 |     3 +
 checkpoint-7000/scheduler.pt                  |     3 +
 checkpoint-7000/special_tokens_map.json       |    34 +
 checkpoint-7000/tokenizer.json                |     3 +
 checkpoint-7000/tokenizer.model               |     3 +
 checkpoint-7000/tokenizer_config.json         |  2013 +++
 checkpoint-7000/trainer_state.json            |  8574 +++++++++++
 checkpoint-7000/training_args.bin             |     3 +
 checkpoint-7500/chat_template.jinja           |     4 +
 checkpoint-7500/config.json                   |    77 +
 checkpoint-7500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-7500/model.safetensors.index.json  |   472 +
 checkpoint-7500/optimizer.pt                  |     3 +
 checkpoint-7500/rng_state.pth                 |     3 +
 checkpoint-7500/scheduler.pt                  |     3 +
 checkpoint-7500/special_tokens_map.json       |    34 +
 checkpoint-7500/tokenizer.json                |     3 +
 checkpoint-7500/tokenizer.model               |     3 +
 checkpoint-7500/tokenizer_config.json         |  2013 +++
 checkpoint-7500/trainer_state.json            |  9184 ++++++++++++
 checkpoint-7500/training_args.bin             |     3 +
 checkpoint-8000/chat_template.jinja           |     4 +
 checkpoint-8000/config.json                   |    77 +
 checkpoint-8000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-8000/model.safetensors.index.json  |   472 +
 checkpoint-8000/optimizer.pt                  |     3 +
 checkpoint-8000/rng_state.pth                 |     3 +
 checkpoint-8000/scheduler.pt                  |     3 +
 checkpoint-8000/special_tokens_map.json       |    34 +
 checkpoint-8000/tokenizer.json                |     3 +
 checkpoint-8000/tokenizer.model               |     3 +
 checkpoint-8000/tokenizer_config.json         |  2013 +++
 checkpoint-8000/trainer_state.json            |  9794 +++++++++++++
 checkpoint-8000/training_args.bin             |     3 +
 checkpoint-8500/chat_template.jinja           |     4 +
 checkpoint-8500/config.json                   |    77 +
 checkpoint-8500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-8500/model.safetensors.index.json  |   472 +
 checkpoint-8500/optimizer.pt                  |     3 +
 checkpoint-8500/rng_state.pth                 |     3 +
 checkpoint-8500/scheduler.pt                  |     3 +
 checkpoint-8500/special_tokens_map.json       |    34 +
 checkpoint-8500/tokenizer.json                |     3 +
 checkpoint-8500/tokenizer.model               |     3 +
 checkpoint-8500/tokenizer_config.json         |  2013 +++
 checkpoint-8500/trainer_state.json            | 10404 +++++++++++++
 checkpoint-8500/training_args.bin             |     3 +
 checkpoint-9000/chat_template.jinja           |     4 +
 checkpoint-9000/config.json                   |    77 +
 checkpoint-9000/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-9000/model.safetensors.index.json  |   472 +
 checkpoint-9000/optimizer.pt                  |     3 +
 checkpoint-9000/rng_state.pth                 |     3 +
 checkpoint-9000/scheduler.pt                  |     3 +
 checkpoint-9000/special_tokens_map.json       |    34 +
 checkpoint-9000/tokenizer.json                |     3 +
 checkpoint-9000/tokenizer.model               |     3 +
 checkpoint-9000/tokenizer_config.json         |  2013 +++
 checkpoint-9000/trainer_state.json            | 11014 ++++++++++++++
 checkpoint-9000/training_args.bin             |     3 +
 checkpoint-9500/chat_template.jinja           |     4 +
 checkpoint-9500/config.json                   |    77 +
 checkpoint-9500/generation_config.json        |     9 +
 .../model-00001-of-00004.safetensors          |     3 +
 .../model-00002-of-00004.safetensors          |     3 +
 .../model-00003-of-00004.safetensors          |     3 +
 .../model-00004-of-00004.safetensors          |     3 +
 checkpoint-9500/model.safetensors.index.json  |   472 +
 checkpoint-9500/optimizer.pt                  |     3 +
 checkpoint-9500/rng_state.pth                 |     3 +
 checkpoint-9500/scheduler.pt                  |     3 +
 checkpoint-9500/special_tokens_map.json       |    34 +
 checkpoint-9500/tokenizer.json                |     3 +
 checkpoint-9500/tokenizer.model               |     3 +
 checkpoint-9500/tokenizer_config.json         |  2013 +++
 checkpoint-9500/trainer_state.json            | 11624 +++++++++++++++
 checkpoint-9500/training_args.bin             |     3 +
 config.json                                   |    77 +
 generation_config.json                        |     9 +
 model-00001-of-00004.safetensors              |     3 +
 model-00002-of-00004.safetensors              |     3 +
 model-00003-of-00004.safetensors              |     3 +
 model-00004-of-00004.safetensors              |     3 +
 model.safetensors.index.json                  |   472 +
 special_tokens_map.json                       |    34 +
 tokenizer.json                                |     3 +
 tokenizer.model                               |     3 +
 tokenizer_config.json                         |  2013 +++
 354 files changed, 184444 insertions(+)
 create mode 100644 .gitattributes
 create mode 100644 README.md
 create mode 100644 chat_template.jinja
 create mode 100644 checkpoint-1000/chat_template.jinja
 create mode 100644 checkpoint-1000/config.json
 create mode 100644 checkpoint-1000/generation_config.json
 create mode 100644 checkpoint-1000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-1000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-1000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-1000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-1000/model.safetensors.index.json
 create mode 100644 checkpoint-1000/optimizer.pt
 create mode 100644 checkpoint-1000/rng_state.pth
 create mode 100644 checkpoint-1000/scheduler.pt
 create mode 100644 checkpoint-1000/special_tokens_map.json
 create mode 100644 checkpoint-1000/tokenizer.json
 create mode 100644 checkpoint-1000/tokenizer.model
 create mode 100644 checkpoint-1000/tokenizer_config.json
 create mode 100644 checkpoint-1000/trainer_state.json
 create mode 100644 checkpoint-1000/training_args.bin
 create mode 100644 checkpoint-10000/chat_template.jinja
 create mode 100644 checkpoint-10000/config.json
 create mode 100644 checkpoint-10000/generation_config.json
 create mode 100644 checkpoint-10000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-10000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-10000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-10000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-10000/model.safetensors.index.json
 create mode 100644 checkpoint-10000/optimizer.pt
 create mode 100644 checkpoint-10000/rng_state.pth
 create mode 100644 checkpoint-10000/scheduler.pt
 create mode 100644 checkpoint-10000/special_tokens_map.json
 create mode 100644 checkpoint-10000/tokenizer.json
 create mode 100644 checkpoint-10000/tokenizer.model
 create mode 100644 checkpoint-10000/tokenizer_config.json
 create mode 100644 checkpoint-10000/trainer_state.json
 create mode 100644 checkpoint-10000/training_args.bin
 create mode 100644 checkpoint-1500/chat_template.jinja
 create mode 100644 checkpoint-1500/config.json
 create mode 100644 checkpoint-1500/generation_config.json
 create mode 100644 checkpoint-1500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-1500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-1500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-1500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-1500/model.safetensors.index.json
 create mode 100644 checkpoint-1500/optimizer.pt
 create mode 100644 checkpoint-1500/rng_state.pth
 create mode 100644 checkpoint-1500/scheduler.pt
 create mode 100644 checkpoint-1500/special_tokens_map.json
 create mode 100644 checkpoint-1500/tokenizer.json
 create mode 100644 checkpoint-1500/tokenizer.model
 create mode 100644 checkpoint-1500/tokenizer_config.json
 create mode 100644 checkpoint-1500/trainer_state.json
 create mode 100644 checkpoint-1500/training_args.bin
 create mode 100644 checkpoint-2000/chat_template.jinja
 create mode 100644 checkpoint-2000/config.json
 create mode 100644 checkpoint-2000/generation_config.json
 create mode 100644 checkpoint-2000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-2000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-2000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-2000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-2000/model.safetensors.index.json
 create mode 100644 checkpoint-2000/optimizer.pt
 create mode 100644 checkpoint-2000/rng_state.pth
 create mode 100644 checkpoint-2000/scheduler.pt
 create mode 100644 checkpoint-2000/special_tokens_map.json
 create mode 100644 checkpoint-2000/tokenizer.json
 create mode 100644 checkpoint-2000/tokenizer.model
 create mode 100644 checkpoint-2000/tokenizer_config.json
 create mode 100644 checkpoint-2000/trainer_state.json
 create mode 100644 checkpoint-2000/training_args.bin
 create mode 100644 checkpoint-2500/chat_template.jinja
 create mode 100644 checkpoint-2500/config.json
 create mode 100644 checkpoint-2500/generation_config.json
 create mode 100644 checkpoint-2500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-2500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-2500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-2500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-2500/model.safetensors.index.json
 create mode 100644 checkpoint-2500/optimizer.pt
 create mode 100644 checkpoint-2500/rng_state.pth
 create mode 100644 checkpoint-2500/scheduler.pt
 create mode 100644 checkpoint-2500/special_tokens_map.json
 create mode 100644 checkpoint-2500/tokenizer.json
 create mode 100644 checkpoint-2500/tokenizer.model
 create mode 100644 checkpoint-2500/tokenizer_config.json
 create mode 100644 checkpoint-2500/trainer_state.json
 create mode 100644 checkpoint-2500/training_args.bin
 create mode 100644 checkpoint-3000/chat_template.jinja
 create mode 100644 checkpoint-3000/config.json
 create mode 100644 checkpoint-3000/generation_config.json
 create mode 100644 checkpoint-3000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-3000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-3000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-3000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-3000/model.safetensors.index.json
 create mode 100644 checkpoint-3000/optimizer.pt
 create mode 100644 checkpoint-3000/rng_state.pth
 create mode 100644 checkpoint-3000/scheduler.pt
 create mode 100644 checkpoint-3000/special_tokens_map.json
 create mode 100644 checkpoint-3000/tokenizer.json
 create mode 100644 checkpoint-3000/tokenizer.model
 create mode 100644 checkpoint-3000/tokenizer_config.json
 create mode 100644 checkpoint-3000/trainer_state.json
 create mode 100644 checkpoint-3000/training_args.bin
 create mode 100644 checkpoint-3500/chat_template.jinja
 create mode 100644 checkpoint-3500/config.json
 create mode 100644 checkpoint-3500/generation_config.json
 create mode 100644 checkpoint-3500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-3500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-3500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-3500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-3500/model.safetensors.index.json
 create mode 100644 checkpoint-3500/optimizer.pt
 create mode 100644 checkpoint-3500/rng_state.pth
 create mode 100644 checkpoint-3500/scheduler.pt
 create mode 100644 checkpoint-3500/special_tokens_map.json
 create mode 100644 checkpoint-3500/tokenizer.json
 create mode 100644 checkpoint-3500/tokenizer.model
 create mode 100644 checkpoint-3500/tokenizer_config.json
 create mode 100644 checkpoint-3500/trainer_state.json
 create mode 100644 checkpoint-3500/training_args.bin
 create mode 100644 checkpoint-4000/chat_template.jinja
 create mode 100644 checkpoint-4000/config.json
 create mode 100644 checkpoint-4000/generation_config.json
 create mode 100644 checkpoint-4000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-4000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-4000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-4000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-4000/model.safetensors.index.json
 create mode 100644 checkpoint-4000/optimizer.pt
 create mode 100644 checkpoint-4000/rng_state.pth
 create mode 100644 checkpoint-4000/scheduler.pt
 create mode 100644 checkpoint-4000/special_tokens_map.json
 create mode 100644 checkpoint-4000/tokenizer.json
 create mode 100644 checkpoint-4000/tokenizer.model
 create mode 100644 checkpoint-4000/tokenizer_config.json
 create mode 100644 checkpoint-4000/trainer_state.json
 create mode 100644 checkpoint-4000/training_args.bin
 create mode 100644 checkpoint-4500/chat_template.jinja
 create mode 100644 checkpoint-4500/config.json
 create mode 100644 checkpoint-4500/generation_config.json
 create mode 100644 checkpoint-4500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-4500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-4500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-4500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-4500/model.safetensors.index.json
 create mode 100644 checkpoint-4500/optimizer.pt
 create mode 100644 checkpoint-4500/rng_state.pth
 create mode 100644 checkpoint-4500/scheduler.pt
 create mode 100644 checkpoint-4500/special_tokens_map.json
 create mode 100644 checkpoint-4500/tokenizer.json
 create mode 100644 checkpoint-4500/tokenizer.model
 create mode 100644 checkpoint-4500/tokenizer_config.json
 create mode 100644 checkpoint-4500/trainer_state.json
 create mode 100644 checkpoint-4500/training_args.bin
 create mode 100644 checkpoint-500/chat_template.jinja
 create mode 100644 checkpoint-500/config.json
 create mode 100644 checkpoint-500/generation_config.json
 create mode 100644 checkpoint-500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-500/model.safetensors.index.json
 create mode 100644 checkpoint-500/optimizer.pt
 create mode 100644 checkpoint-500/rng_state.pth
 create mode 100644 checkpoint-500/scheduler.pt
 create mode 100644 checkpoint-500/special_tokens_map.json
 create mode 100644 checkpoint-500/tokenizer.json
 create mode 100644 checkpoint-500/tokenizer.model
 create mode 100644 checkpoint-500/tokenizer_config.json
 create mode 100644 checkpoint-500/trainer_state.json
 create mode 100644 checkpoint-500/training_args.bin
 create mode 100644 checkpoint-5000/chat_template.jinja
 create mode 100644 checkpoint-5000/config.json
 create mode 100644 checkpoint-5000/generation_config.json
 create mode 100644 checkpoint-5000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-5000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-5000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-5000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-5000/model.safetensors.index.json
 create mode 100644 checkpoint-5000/optimizer.pt
 create mode 100644 checkpoint-5000/rng_state.pth
 create mode 100644 checkpoint-5000/scheduler.pt
 create mode 100644 checkpoint-5000/special_tokens_map.json
 create mode 100644 checkpoint-5000/tokenizer.json
 create mode 100644 checkpoint-5000/tokenizer.model
 create mode 100644 checkpoint-5000/tokenizer_config.json
 create mode 100644 checkpoint-5000/trainer_state.json
 create mode 100644 checkpoint-5000/training_args.bin
 create mode 100644 checkpoint-5500/chat_template.jinja
 create mode 100644 checkpoint-5500/config.json
 create mode 100644 checkpoint-5500/generation_config.json
 create mode 100644 checkpoint-5500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-5500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-5500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-5500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-5500/model.safetensors.index.json
 create mode 100644 checkpoint-5500/optimizer.pt
 create mode 100644 checkpoint-5500/rng_state.pth
 create mode 100644 checkpoint-5500/scheduler.pt
 create mode 100644 checkpoint-5500/special_tokens_map.json
 create mode 100644 checkpoint-5500/tokenizer.json
 create mode 100644 checkpoint-5500/tokenizer.model
 create mode 100644 checkpoint-5500/tokenizer_config.json
 create mode 100644 checkpoint-5500/trainer_state.json
 create mode 100644 checkpoint-5500/training_args.bin
 create mode 100644 checkpoint-6000/chat_template.jinja
 create mode 100644 checkpoint-6000/config.json
 create mode 100644 checkpoint-6000/generation_config.json
 create mode 100644 checkpoint-6000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-6000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-6000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-6000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-6000/model.safetensors.index.json
 create mode 100644 checkpoint-6000/optimizer.pt
 create mode 100644 checkpoint-6000/rng_state.pth
 create mode 100644 checkpoint-6000/scheduler.pt
 create mode 100644 checkpoint-6000/special_tokens_map.json
 create mode 100644 checkpoint-6000/tokenizer.json
 create mode 100644 checkpoint-6000/tokenizer.model
 create mode 100644 checkpoint-6000/tokenizer_config.json
 create mode 100644 checkpoint-6000/trainer_state.json
 create mode 100644 checkpoint-6000/training_args.bin
 create mode 100644 checkpoint-6500/chat_template.jinja
 create mode 100644 checkpoint-6500/config.json
 create mode 100644 checkpoint-6500/generation_config.json
 create mode 100644 checkpoint-6500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-6500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-6500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-6500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-6500/model.safetensors.index.json
 create mode 100644 checkpoint-6500/optimizer.pt
 create mode 100644 checkpoint-6500/rng_state.pth
 create mode 100644 checkpoint-6500/scheduler.pt
 create mode 100644 checkpoint-6500/special_tokens_map.json
 create mode 100644 checkpoint-6500/tokenizer.json
 create mode 100644 checkpoint-6500/tokenizer.model
 create mode 100644 checkpoint-6500/tokenizer_config.json
 create mode 100644 checkpoint-6500/trainer_state.json
 create mode 100644 checkpoint-6500/training_args.bin
 create mode 100644 checkpoint-7000/chat_template.jinja
 create mode 100644 checkpoint-7000/config.json
 create mode 100644 checkpoint-7000/generation_config.json
 create mode 100644 checkpoint-7000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-7000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-7000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-7000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-7000/model.safetensors.index.json
 create mode 100644 checkpoint-7000/optimizer.pt
 create mode 100644 checkpoint-7000/rng_state.pth
 create mode 100644 checkpoint-7000/scheduler.pt
 create mode 100644 checkpoint-7000/special_tokens_map.json
 create mode 100644 checkpoint-7000/tokenizer.json
 create mode 100644 checkpoint-7000/tokenizer.model
 create mode 100644 checkpoint-7000/tokenizer_config.json
 create mode 100644 checkpoint-7000/trainer_state.json
 create mode 100644 checkpoint-7000/training_args.bin
 create mode 100644 checkpoint-7500/chat_template.jinja
 create mode 100644 checkpoint-7500/config.json
 create mode 100644 checkpoint-7500/generation_config.json
 create mode 100644 checkpoint-7500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-7500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-7500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-7500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-7500/model.safetensors.index.json
 create mode 100644 checkpoint-7500/optimizer.pt
 create mode 100644 checkpoint-7500/rng_state.pth
 create mode 100644 checkpoint-7500/scheduler.pt
 create mode 100644 checkpoint-7500/special_tokens_map.json
 create mode 100644 checkpoint-7500/tokenizer.json
 create mode 100644 checkpoint-7500/tokenizer.model
 create mode 100644 checkpoint-7500/tokenizer_config.json
 create mode 100644 checkpoint-7500/trainer_state.json
 create mode 100644 checkpoint-7500/training_args.bin
 create mode 100644 checkpoint-8000/chat_template.jinja
 create mode 100644 checkpoint-8000/config.json
 create mode 100644 checkpoint-8000/generation_config.json
 create mode 100644 checkpoint-8000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-8000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-8000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-8000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-8000/model.safetensors.index.json
 create mode 100644 checkpoint-8000/optimizer.pt
 create mode 100644 checkpoint-8000/rng_state.pth
 create mode 100644 checkpoint-8000/scheduler.pt
 create mode 100644 checkpoint-8000/special_tokens_map.json
 create mode 100644 checkpoint-8000/tokenizer.json
 create mode 100644 checkpoint-8000/tokenizer.model
 create mode 100644 checkpoint-8000/tokenizer_config.json
 create mode 100644 checkpoint-8000/trainer_state.json
 create mode 100644 checkpoint-8000/training_args.bin
 create mode 100644 checkpoint-8500/chat_template.jinja
 create mode 100644 checkpoint-8500/config.json
 create mode 100644 checkpoint-8500/generation_config.json
 create mode 100644 checkpoint-8500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-8500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-8500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-8500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-8500/model.safetensors.index.json
 create mode 100644 checkpoint-8500/optimizer.pt
 create mode 100644 checkpoint-8500/rng_state.pth
 create mode 100644 checkpoint-8500/scheduler.pt
 create mode 100644 checkpoint-8500/special_tokens_map.json
 create mode 100644 checkpoint-8500/tokenizer.json
 create mode 100644 checkpoint-8500/tokenizer.model
 create mode 100644 checkpoint-8500/tokenizer_config.json
 create mode 100644 checkpoint-8500/trainer_state.json
 create mode 100644 checkpoint-8500/training_args.bin
 create mode 100644 checkpoint-9000/chat_template.jinja
 create mode 100644 checkpoint-9000/config.json
 create mode 100644 checkpoint-9000/generation_config.json
 create mode 100644 checkpoint-9000/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-9000/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-9000/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-9000/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-9000/model.safetensors.index.json
 create mode 100644 checkpoint-9000/optimizer.pt
 create mode 100644 checkpoint-9000/rng_state.pth
 create mode 100644 checkpoint-9000/scheduler.pt
 create mode 100644 checkpoint-9000/special_tokens_map.json
 create mode 100644 checkpoint-9000/tokenizer.json
 create mode 100644 checkpoint-9000/tokenizer.model
 create mode 100644 checkpoint-9000/tokenizer_config.json
 create mode 100644 checkpoint-9000/trainer_state.json
 create mode 100644 checkpoint-9000/training_args.bin
 create mode 100644 checkpoint-9500/chat_template.jinja
 create mode 100644 checkpoint-9500/config.json
 create mode 100644 checkpoint-9500/generation_config.json
 create mode 100644 checkpoint-9500/model-00001-of-00004.safetensors
 create mode 100644 checkpoint-9500/model-00002-of-00004.safetensors
 create mode 100644 checkpoint-9500/model-00003-of-00004.safetensors
 create mode 100644 checkpoint-9500/model-00004-of-00004.safetensors
 create mode 100644 checkpoint-9500/model.safetensors.index.json
 create mode 100644 checkpoint-9500/optimizer.pt
 create mode 100644 checkpoint-9500/rng_state.pth
 create mode 100644 checkpoint-9500/scheduler.pt
 create mode 100644 checkpoint-9500/special_tokens_map.json
 create mode 100644 checkpoint-9500/tokenizer.json
 create mode 100644 checkpoint-9500/tokenizer.model
 create mode 100644 checkpoint-9500/tokenizer_config.json
 create mode 100644 checkpoint-9500/trainer_state.json
 create mode 100644 checkpoint-9500/training_args.bin
 create mode 100644 config.json
 create mode 100644 generation_config.json
 create mode 100644 model-00001-of-00004.safetensors
 create mode 100644 model-00002-of-00004.safetensors
 create mode 100644 model-00003-of-00004.safetensors
 create mode 100644 model-00004-of-00004.safetensors
 create mode 100644 model.safetensors.index.json
 create mode 100644 special_tokens_map.json
 create mode 100644 tokenizer.json
 create mode 100644 tokenizer.model
 create mode 100644 tokenizer_config.json

diff --git a/.gitattributes b/.gitattributes
new file mode 100644
index 0000000..1858320
--- /dev/null
+++ b/.gitattributes
@@ -0,0 +1,56 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+checkpoint-500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-1000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-1500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-2000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-2500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-3000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-3500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-4000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-4500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-5000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-5500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-6000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-6500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-7000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-7500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-8000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-8500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-9000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-9500/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-10000/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..2480968
--- /dev/null
+++ b/README.md
@@ -0,0 +1,201 @@
+---
+library_name: transformers
+tags:
+- trl
+- sft
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
\ No newline at end of file
diff --git a/chat_template.jinja b/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-1000/chat_template.jinja b/checkpoint-1000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-1000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-1000/config.json b/checkpoint-1000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-1000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-1000/generation_config.json b/checkpoint-1000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-1000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-1000/model-00001-of-00004.safetensors b/checkpoint-1000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-1000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-1000/model-00002-of-00004.safetensors b/checkpoint-1000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-1000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-1000/model-00003-of-00004.safetensors b/checkpoint-1000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..aa91ab4
--- /dev/null
+++ b/checkpoint-1000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f0e13e9014c8be4dac2d64d118652a4b3f84d4002ca782dab7ce59a003359bc5
+size 4962221464
diff --git a/checkpoint-1000/model-00004-of-00004.safetensors b/checkpoint-1000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..0bb7911
--- /dev/null
+++ b/checkpoint-1000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7e00359ead0d8dca629034fc5b55db125a9fc50d8856d43c1ef33a6d4521d33
+size 3670322200
diff --git a/checkpoint-1000/model.safetensors.index.json b/checkpoint-1000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-1000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-1000/optimizer.pt b/checkpoint-1000/optimizer.pt
new file mode 100644
index 0000000..bb0737d
--- /dev/null
+++ b/checkpoint-1000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a61e2932857e62e09d0cef7a82f9360ccd9d3b0d428e1d24a78eac70d1cb7aaa
+size 7927918951
diff --git a/checkpoint-1000/rng_state.pth b/checkpoint-1000/rng_state.pth
new file mode 100644
index 0000000..674db9d
--- /dev/null
+++ b/checkpoint-1000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6e9c81fd74b2d0afb9f476e77fa89b23bd54271539d11a8d1c13b69a3096b56
+size 14645
diff --git a/checkpoint-1000/scheduler.pt b/checkpoint-1000/scheduler.pt
new file mode 100644
index 0000000..64d9870
--- /dev/null
+++ b/checkpoint-1000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6078e52773a379f026b8fbbbc5546ab7ed9418d05b713f8aff2a4f3c7e12f108
+size 1465
diff --git a/checkpoint-1000/special_tokens_map.json b/checkpoint-1000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-1000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-1000/tokenizer.json b/checkpoint-1000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-1000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-1000/tokenizer.model b/checkpoint-1000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-1000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-1000/tokenizer_config.json b/checkpoint-1000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-1000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-1000/trainer_state.json b/checkpoint-1000/trainer_state.json
new file mode 100644
index 0000000..38b6631
--- /dev/null
+++ b/checkpoint-1000/trainer_state.json
@@ -0,0 +1,1254 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.6,
+  "eval_steps": 100,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1731341502193254e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-1000/training_args.bin b/checkpoint-1000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-1000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-10000/chat_template.jinja b/checkpoint-10000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-10000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-10000/config.json b/checkpoint-10000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-10000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-10000/generation_config.json b/checkpoint-10000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-10000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-10000/model-00001-of-00004.safetensors b/checkpoint-10000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-10000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-10000/model-00002-of-00004.safetensors b/checkpoint-10000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-10000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-10000/model-00003-of-00004.safetensors b/checkpoint-10000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..48ad5e5
--- /dev/null
+++ b/checkpoint-10000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87ed143571b20e30b1af91677aa3a0efc1b94fc7a6bc3a66f764ad27ab898d4b
+size 4962221464
diff --git a/checkpoint-10000/model-00004-of-00004.safetensors b/checkpoint-10000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..f793b72
--- /dev/null
+++ b/checkpoint-10000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cced0983c4c03fb39838a94e571ba14bf4058058cdc63d741cc9aa4156d810fe
+size 3670322200
diff --git a/checkpoint-10000/model.safetensors.index.json b/checkpoint-10000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-10000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-10000/optimizer.pt b/checkpoint-10000/optimizer.pt
new file mode 100644
index 0000000..a60c4ef
--- /dev/null
+++ b/checkpoint-10000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7dc85ed31c7a41ca7214a379404edcc7b17b647b08341ca780c40676a99fd64d
+size 7927918951
diff --git a/checkpoint-10000/rng_state.pth b/checkpoint-10000/rng_state.pth
new file mode 100644
index 0000000..ab0fbb0
--- /dev/null
+++ b/checkpoint-10000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed1ee220e360121191f2ae3c5239d1f147fbfe00f61b820afa469187586ca3ad
+size 14645
diff --git a/checkpoint-10000/scheduler.pt b/checkpoint-10000/scheduler.pt
new file mode 100644
index 0000000..3f985b3
--- /dev/null
+++ b/checkpoint-10000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99613dd40529fbecd78704a9a6d56d37a9ef57f01e576dc59447aa1418051563
+size 1465
diff --git a/checkpoint-10000/special_tokens_map.json b/checkpoint-10000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-10000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-10000/tokenizer.json b/checkpoint-10000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-10000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-10000/tokenizer.model b/checkpoint-10000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-10000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-10000/tokenizer_config.json b/checkpoint-10000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-10000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-10000/trainer_state.json b/checkpoint-10000/trainer_state.json
new file mode 100644
index 0000000..9dd03e3
--- /dev/null
+++ b/checkpoint-10000/trainer_state.json
@@ -0,0 +1,12234 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 16.0,
+  "eval_steps": 100,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    },
+    {
+      "entropy": 0.053174392925575374,
+      "epoch": 12.016,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.535555555555556e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905574269592762,
+      "num_tokens": 214920834.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.05121934600174427,
+      "epoch": 12.032,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.513333333333334e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907024279236794,
+      "num_tokens": 215207212.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.05282459647860378,
+      "epoch": 12.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.491111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904284842312336,
+      "num_tokens": 215491551.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.0523437969153747,
+      "epoch": 12.064,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.468888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907445590943098,
+      "num_tokens": 215773788.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.05109455803176388,
+      "epoch": 12.08,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990898734331131,
+      "num_tokens": 216063342.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.052485697087831795,
+      "epoch": 12.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.424444444444444e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905819371342659,
+      "num_tokens": 216346341.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.05154973501339555,
+      "epoch": 12.112,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.402222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906920839101077,
+      "num_tokens": 216631586.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.05027051754295826,
+      "epoch": 12.128,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908485073596239,
+      "num_tokens": 216921387.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.051271379156969485,
+      "epoch": 12.144,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.357777777777779e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907938674092293,
+      "num_tokens": 217208948.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.05280535111669451,
+      "epoch": 12.16,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904177527874708,
+      "num_tokens": 217488087.0,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_math_entropy": 0.17994125121831894,
+      "eval_math_loss": 1.3322224617004395,
+      "eval_math_mean_token_accuracy": 0.8160332980155944,
+      "eval_math_num_tokens": 217488087.0,
+      "eval_math_runtime": 35.8103,
+      "eval_math_samples_per_second": 13.962,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_physics_entropy": 0.20668217623233795,
+      "eval_physics_loss": 1.3987233638763428,
+      "eval_physics_mean_token_accuracy": 0.7963971982002258,
+      "eval_physics_num_tokens": 217488087.0,
+      "eval_physics_runtime": 41.4958,
+      "eval_physics_samples_per_second": 12.049,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7600
+    },
+    {
+      "entropy": 0.05196192401926965,
+      "epoch": 12.176,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.313333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906202860176563,
+      "num_tokens": 217770038.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.051902253658045086,
+      "epoch": 12.192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.2911111111111115e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9906028714030981,
+      "num_tokens": 218050486.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.052790193830151114,
+      "epoch": 12.208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.268888888888889e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904781151562929,
+      "num_tokens": 218331465.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.052298348920885473,
+      "epoch": 12.224,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.246666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906631808727979,
+      "num_tokens": 218618801.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.050798607314936815,
+      "epoch": 12.24,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9909202449023724,
+      "num_tokens": 218910596.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.052114113536663356,
+      "epoch": 12.256,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.202222222222223e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905635304749012,
+      "num_tokens": 219194842.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.05205410345224663,
+      "epoch": 12.272,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990602058544755,
+      "num_tokens": 219484076.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.049987882829736915,
+      "epoch": 12.288,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.157777777777778e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909869834780693,
+      "num_tokens": 219780222.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.05258604409173131,
+      "epoch": 12.304,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.135555555555556e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904522281140089,
+      "num_tokens": 220065272.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.05357218148419633,
+      "epoch": 12.32,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9902280792593956,
+      "num_tokens": 220344110.0,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_math_entropy": 0.1793700819015503,
+      "eval_math_loss": 1.3329575061798096,
+      "eval_math_mean_token_accuracy": 0.8161038937568664,
+      "eval_math_num_tokens": 220344110.0,
+      "eval_math_runtime": 35.8174,
+      "eval_math_samples_per_second": 13.96,
+      "eval_math_steps_per_second": 3.49,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_physics_entropy": 0.20628326386213303,
+      "eval_physics_loss": 1.3988101482391357,
+      "eval_physics_mean_token_accuracy": 0.7963225340843201,
+      "eval_physics_num_tokens": 220344110.0,
+      "eval_physics_runtime": 41.5047,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7700
+    },
+    {
+      "entropy": 0.05083432825049385,
+      "epoch": 12.336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.091111111111111e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906401328742505,
+      "num_tokens": 220632516.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.05188987164292484,
+      "epoch": 12.352,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.06888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905499298125505,
+      "num_tokens": 220918123.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.05209730467759073,
+      "epoch": 12.368,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.046666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905961699783802,
+      "num_tokens": 221206063.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.05238552400842309,
+      "epoch": 12.384,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.024444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904579646885395,
+      "num_tokens": 221488317.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.05181863873731345,
+      "epoch": 12.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905485171824694,
+      "num_tokens": 221776008.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.05192954689264297,
+      "epoch": 12.416,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904752410948277,
+      "num_tokens": 222057789.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.052794202556833626,
+      "epoch": 12.432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.957777777777778e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990259325504303,
+      "num_tokens": 222339915.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.052391084167174994,
+      "epoch": 12.448,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903523031622171,
+      "num_tokens": 222619687.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.05029154985677451,
+      "epoch": 12.464,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.9133333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906887255609036,
+      "num_tokens": 222913264.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.05082799810916185,
+      "epoch": 12.48,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906626105308532,
+      "num_tokens": 223202947.0,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_math_entropy": 0.180211255133152,
+      "eval_math_loss": 1.333121657371521,
+      "eval_math_mean_token_accuracy": 0.8160158872604371,
+      "eval_math_num_tokens": 223202947.0,
+      "eval_math_runtime": 35.8241,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_physics_entropy": 0.20699969559907913,
+      "eval_physics_loss": 1.3986775875091553,
+      "eval_physics_mean_token_accuracy": 0.7963728904724121,
+      "eval_physics_num_tokens": 223202947.0,
+      "eval_physics_runtime": 41.4724,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 7800
+    },
+    {
+      "entropy": 0.05148934966418892,
+      "epoch": 12.496,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.8688888888888895e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905773665755987,
+      "num_tokens": 223491040.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.05243352035759017,
+      "epoch": 12.512,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.846666666666667e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903561118990183,
+      "num_tokens": 223775825.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.05058162130881101,
+      "epoch": 12.528,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.824444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990830871462822,
+      "num_tokens": 224068092.0,
+      "step": 7830
+    },
+    {
+      "entropy": 0.051722967089153825,
+      "epoch": 12.544,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.802222222222222e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903257399797439,
+      "num_tokens": 224354624.0,
+      "step": 7840
+    },
+    {
+      "entropy": 0.05123099538031965,
+      "epoch": 12.56,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906783111393451,
+      "num_tokens": 224641974.0,
+      "step": 7850
+    },
+    {
+      "entropy": 0.05136660183779895,
+      "epoch": 12.576,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.7577777777777784e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906629983335733,
+      "num_tokens": 224932014.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.05145444863010198,
+      "epoch": 12.592,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.735555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9903172593563795,
+      "num_tokens": 225222434.0,
+      "step": 7870
+    },
+    {
+      "entropy": 0.05075638788985089,
+      "epoch": 12.608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.713333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906791742891073,
+      "num_tokens": 225515803.0,
+      "step": 7880
+    },
+    {
+      "entropy": 0.05076117676217109,
+      "epoch": 12.624,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.691111111111111e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906085334718228,
+      "num_tokens": 225810131.0,
+      "step": 7890
+    },
+    {
+      "entropy": 0.05141493146074936,
+      "epoch": 12.64,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904979523271322,
+      "num_tokens": 226100172.0,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_math_entropy": 0.17927622705698013,
+      "eval_math_loss": 1.332814335823059,
+      "eval_math_mean_token_accuracy": 0.8159863357543945,
+      "eval_math_num_tokens": 226100172.0,
+      "eval_math_runtime": 35.8058,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_physics_entropy": 0.20605486339330673,
+      "eval_physics_loss": 1.3984746932983398,
+      "eval_physics_mean_token_accuracy": 0.7964399604797363,
+      "eval_physics_num_tokens": 226100172.0,
+      "eval_physics_runtime": 41.5077,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7900
+    },
+    {
+      "entropy": 0.052684175770264116,
+      "epoch": 12.656,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.646666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902132380753755,
+      "num_tokens": 226379811.0,
+      "step": 7910
+    },
+    {
+      "entropy": 0.05348572533112019,
+      "epoch": 12.672,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.990258751064539,
+      "num_tokens": 226662085.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.05106104960432276,
+      "epoch": 12.688,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.602222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9902874868363142,
+      "num_tokens": 226950257.0,
+      "step": 7930
+    },
+    {
+      "entropy": 0.0518765551969409,
+      "epoch": 12.704,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905543696135283,
+      "num_tokens": 227238669.0,
+      "step": 7940
+    },
+    {
+      "entropy": 0.05253795627504587,
+      "epoch": 12.72,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990134946629405,
+      "num_tokens": 227518695.0,
+      "step": 7950
+    },
+    {
+      "entropy": 0.05207758563337848,
+      "epoch": 12.736,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.535555555555555e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990263070166111,
+      "num_tokens": 227806156.0,
+      "step": 7960
+    },
+    {
+      "entropy": 0.05179664236493409,
+      "epoch": 12.752,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.513333333333333e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904587935656309,
+      "num_tokens": 228091641.0,
+      "step": 7970
+    },
+    {
+      "entropy": 0.05190181992948055,
+      "epoch": 12.768,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.4911111111111115e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9901351545006036,
+      "num_tokens": 228376952.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.0528141722548753,
+      "epoch": 12.784,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.468888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901187218725681,
+      "num_tokens": 228657003.0,
+      "step": 7990
+    },
+    {
+      "entropy": 0.051391151722054926,
+      "epoch": 12.8,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.446666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9902074325829744,
+      "num_tokens": 228942799.0,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_math_entropy": 0.1799713587164879,
+      "eval_math_loss": 1.332999587059021,
+      "eval_math_mean_token_accuracy": 0.8159932594299316,
+      "eval_math_num_tokens": 228942799.0,
+      "eval_math_runtime": 35.8043,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_physics_entropy": 0.20658580523729325,
+      "eval_physics_loss": 1.3991401195526123,
+      "eval_physics_mean_token_accuracy": 0.7965000495910645,
+      "eval_physics_num_tokens": 228942799.0,
+      "eval_physics_runtime": 41.1662,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 8000
+    },
+    {
+      "entropy": 0.05222373401047662,
+      "epoch": 12.816,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.424444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902573928236962,
+      "num_tokens": 229230267.0,
+      "step": 8010
+    },
+    {
+      "entropy": 0.05221812457311899,
+      "epoch": 12.832,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.402222222222223e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902742177248001,
+      "num_tokens": 229513807.0,
+      "step": 8020
+    },
+    {
+      "entropy": 0.05192019739188254,
+      "epoch": 12.848,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990331218764186,
+      "num_tokens": 229800116.0,
+      "step": 8030
+    },
+    {
+      "entropy": 0.05204707682132721,
+      "epoch": 12.864,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904342792928219,
+      "num_tokens": 230086560.0,
+      "step": 8040
+    },
+    {
+      "entropy": 0.05261244847206399,
+      "epoch": 12.88,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.3355555555555565e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901342626661063,
+      "num_tokens": 230370705.0,
+      "step": 8050
+    },
+    {
+      "entropy": 0.050930758623871954,
+      "epoch": 12.896,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.313333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904414359480143,
+      "num_tokens": 230665090.0,
+      "step": 8060
+    },
+    {
+      "entropy": 0.0521988584427163,
+      "epoch": 12.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.291111111111112e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902524299919605,
+      "num_tokens": 230948774.0,
+      "step": 8070
+    },
+    {
+      "entropy": 0.05130416583269835,
+      "epoch": 12.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.268888888888889e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902212977409363,
+      "num_tokens": 231236122.0,
+      "step": 8080
+    },
+    {
+      "entropy": 0.05124916716013104,
+      "epoch": 12.943999999999999,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.246666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990312222391367,
+      "num_tokens": 231520994.0,
+      "step": 8090
+    },
+    {
+      "entropy": 0.05303966203937307,
+      "epoch": 12.96,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.2244444444444446e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9900147426873446,
+      "num_tokens": 231803509.0,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_math_entropy": 0.18036945784091948,
+      "eval_math_loss": 1.332650899887085,
+      "eval_math_mean_token_accuracy": 0.8161214051246644,
+      "eval_math_num_tokens": 231803509.0,
+      "eval_math_runtime": 35.8002,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_physics_entropy": 0.2067739045023918,
+      "eval_physics_loss": 1.3985861539840698,
+      "eval_physics_mean_token_accuracy": 0.7964418997764587,
+      "eval_physics_num_tokens": 231803509.0,
+      "eval_physics_runtime": 41.4682,
+      "eval_physics_samples_per_second": 12.057,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8100
+    },
+    {
+      "entropy": 0.05145827913656831,
+      "epoch": 12.975999999999999,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.202222222222222e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9905312418937683,
+      "num_tokens": 232095976.0,
+      "step": 8110
+    },
+    {
+      "entropy": 0.05109680832829326,
+      "epoch": 12.992,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903466764837503,
+      "num_tokens": 232382972.0,
+      "step": 8120
+    },
+    {
+      "entropy": 0.0512539794202894,
+      "epoch": 13.008,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.157777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906226065009832,
+      "num_tokens": 232672057.0,
+      "step": 8130
+    },
+    {
+      "entropy": 0.050676750240381804,
+      "epoch": 13.024,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.135555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990827102959156,
+      "num_tokens": 232958227.0,
+      "step": 8140
+    },
+    {
+      "entropy": 0.05172369135543704,
+      "epoch": 13.04,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.1133333333333335e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990977942943573,
+      "num_tokens": 233251931.0,
+      "step": 8150
+    },
+    {
+      "entropy": 0.052580399974249306,
+      "epoch": 13.056,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.091111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905320156365633,
+      "num_tokens": 233531328.0,
+      "step": 8160
+    },
+    {
+      "entropy": 0.05163608979200944,
+      "epoch": 13.072,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.0688888888888896e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990771321207285,
+      "num_tokens": 233816145.0,
+      "step": 8170
+    },
+    {
+      "entropy": 0.05149128795601428,
+      "epoch": 13.088,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.046666666666667e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.990656491741538,
+      "num_tokens": 234100261.0,
+      "step": 8180
+    },
+    {
+      "entropy": 0.0517811676254496,
+      "epoch": 13.104,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.024444444444445e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906980969011784,
+      "num_tokens": 234387753.0,
+      "step": 8190
+    },
+    {
+      "entropy": 0.051599945151247086,
+      "epoch": 13.12,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.002222222222222e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990650337934494,
+      "num_tokens": 234673838.0,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_math_entropy": 0.18022768545150758,
+      "eval_math_loss": 1.3355015516281128,
+      "eval_math_mean_token_accuracy": 0.8160842609405518,
+      "eval_math_num_tokens": 234673838.0,
+      "eval_math_runtime": 35.7873,
+      "eval_math_samples_per_second": 13.971,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_physics_entropy": 0.20687673270702361,
+      "eval_physics_loss": 1.401163935661316,
+      "eval_physics_mean_token_accuracy": 0.7964475116729737,
+      "eval_physics_num_tokens": 234673838.0,
+      "eval_physics_runtime": 41.4759,
+      "eval_physics_samples_per_second": 12.055,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8200
+    },
+    {
+      "entropy": 0.05104603223735467,
+      "epoch": 13.136,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908260378986598,
+      "num_tokens": 234966723.0,
+      "step": 8210
+    },
+    {
+      "entropy": 0.051453781058080496,
+      "epoch": 13.152,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.9577777777777785e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907778527587652,
+      "num_tokens": 235250894.0,
+      "step": 8220
+    },
+    {
+      "entropy": 0.052391895826440306,
+      "epoch": 13.168,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.935555555555556e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904574226588011,
+      "num_tokens": 235538789.0,
+      "step": 8230
+    },
+    {
+      "entropy": 0.05026549928588793,
+      "epoch": 13.184,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.913333333333334e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909756503999233,
+      "num_tokens": 235837152.0,
+      "step": 8240
+    },
+    {
+      "entropy": 0.053065391920972614,
+      "epoch": 13.2,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.891111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901943173259496,
+      "num_tokens": 236116206.0,
+      "step": 8250
+    },
+    {
+      "entropy": 0.051446165784727785,
+      "epoch": 13.216,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.868888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906851042062044,
+      "num_tokens": 236399981.0,
+      "step": 8260
+    },
+    {
+      "entropy": 0.0512137760524638,
+      "epoch": 13.232,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.8466666666666665e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906529590487481,
+      "num_tokens": 236687282.0,
+      "step": 8270
+    },
+    {
+      "entropy": 0.05217738675419241,
+      "epoch": 13.248,
+      "grad_norm": 0.5,
+      "learning_rate": 3.824444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905663672834635,
+      "num_tokens": 236970703.0,
+      "step": 8280
+    },
+    {
+      "entropy": 0.051728677074424925,
+      "epoch": 13.264,
+      "grad_norm": 0.52734375,
+      "learning_rate": 3.8022222222222226e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905138719826937,
+      "num_tokens": 237250995.0,
+      "step": 8290
+    },
+    {
+      "entropy": 0.05051842969842255,
+      "epoch": 13.28,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908084899187088,
+      "num_tokens": 237538556.0,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_math_entropy": 0.1792960543036461,
+      "eval_math_loss": 1.336340308189392,
+      "eval_math_mean_token_accuracy": 0.8159542636871338,
+      "eval_math_num_tokens": 237538556.0,
+      "eval_math_runtime": 35.8026,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_physics_entropy": 0.20614614540338516,
+      "eval_physics_loss": 1.4020493030548096,
+      "eval_physics_mean_token_accuracy": 0.796203773021698,
+      "eval_physics_num_tokens": 237538556.0,
+      "eval_physics_runtime": 41.473,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8300
+    },
+    {
+      "entropy": 0.05342790780123323,
+      "epoch": 13.296,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.757777777777778e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900773618370294,
+      "num_tokens": 237813501.0,
+      "step": 8310
+    },
+    {
+      "entropy": 0.05309537169523537,
+      "epoch": 13.312,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.7355555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903508022427558,
+      "num_tokens": 238094305.0,
+      "step": 8320
+    },
+    {
+      "entropy": 0.05170744662173092,
+      "epoch": 13.328,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.713333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905386842787266,
+      "num_tokens": 238374175.0,
+      "step": 8330
+    },
+    {
+      "entropy": 0.04992100042290985,
+      "epoch": 13.344,
+      "grad_norm": 0.41015625,
+      "learning_rate": 3.6911111111111115e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908522251993418,
+      "num_tokens": 238668031.0,
+      "step": 8340
+    },
+    {
+      "entropy": 0.05189812278840691,
+      "epoch": 13.36,
+      "grad_norm": 0.470703125,
+      "learning_rate": 3.668888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905831549316645,
+      "num_tokens": 238950369.0,
+      "step": 8350
+    },
+    {
+      "entropy": 0.05139565443387255,
+      "epoch": 13.376,
+      "grad_norm": 0.46875,
+      "learning_rate": 3.6466666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906327720731497,
+      "num_tokens": 239233027.0,
+      "step": 8360
+    },
+    {
+      "entropy": 0.05204238067381084,
+      "epoch": 13.392,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.624444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904446199536323,
+      "num_tokens": 239513675.0,
+      "step": 8370
+    },
+    {
+      "entropy": 0.051922239165287466,
+      "epoch": 13.408,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.6022222222222224e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9902938231825829,
+      "num_tokens": 239797199.0,
+      "step": 8380
+    },
+    {
+      "entropy": 0.05086992103606462,
+      "epoch": 13.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905360575765372,
+      "num_tokens": 240082617.0,
+      "step": 8390
+    },
+    {
+      "entropy": 0.052436576853506264,
+      "epoch": 13.44,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.5577777777777785e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903819940984249,
+      "num_tokens": 240365037.0,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_math_entropy": 0.17929287600517274,
+      "eval_math_loss": 1.3361586332321167,
+      "eval_math_mean_token_accuracy": 0.8159851579666137,
+      "eval_math_num_tokens": 240365037.0,
+      "eval_math_runtime": 35.8812,
+      "eval_math_samples_per_second": 13.935,
+      "eval_math_steps_per_second": 3.484,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_physics_entropy": 0.2061038536429405,
+      "eval_physics_loss": 1.4018964767456055,
+      "eval_physics_mean_token_accuracy": 0.7964254455566406,
+      "eval_physics_num_tokens": 240365037.0,
+      "eval_physics_runtime": 41.5825,
+      "eval_physics_samples_per_second": 12.024,
+      "eval_physics_steps_per_second": 3.006,
+      "step": 8400
+    },
+    {
+      "entropy": 0.052103328774683176,
+      "epoch": 13.456,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.535555555555556e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904427032917738,
+      "num_tokens": 240647742.0,
+      "step": 8410
+    },
+    {
+      "entropy": 0.0523280065273866,
+      "epoch": 13.472,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.5133333333333337e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904405385255813,
+      "num_tokens": 240926943.0,
+      "step": 8420
+    },
+    {
+      "entropy": 0.05302672478137538,
+      "epoch": 13.488,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4911111111111113e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903274435549975,
+      "num_tokens": 241206172.0,
+      "step": 8430
+    },
+    {
+      "entropy": 0.05202294115442783,
+      "epoch": 13.504,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4688888888888894e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905512318015098,
+      "num_tokens": 241489988.0,
+      "step": 8440
+    },
+    {
+      "entropy": 0.050564923824276777,
+      "epoch": 13.52,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.446666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907515056431293,
+      "num_tokens": 241781801.0,
+      "step": 8450
+    },
+    {
+      "entropy": 0.05287049149628729,
+      "epoch": 13.536,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.4244444444444446e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903701849281787,
+      "num_tokens": 242064444.0,
+      "step": 8460
+    },
+    {
+      "entropy": 0.05167013480095193,
+      "epoch": 13.552,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.4022222222222222e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904057901352644,
+      "num_tokens": 242354272.0,
+      "step": 8470
+    },
+    {
+      "entropy": 0.0516669807722792,
+      "epoch": 13.568,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9903885431587696,
+      "num_tokens": 242640475.0,
+      "step": 8480
+    },
+    {
+      "entropy": 0.05292565480340272,
+      "epoch": 13.584,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.3577777777777783e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901615660637617,
+      "num_tokens": 242920673.0,
+      "step": 8490
+    },
+    {
+      "entropy": 0.051914384588599205,
+      "epoch": 13.6,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990172453969717,
+      "num_tokens": 243198176.0,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_math_entropy": 0.17909148234128952,
+      "eval_math_loss": 1.3365639448165894,
+      "eval_math_mean_token_accuracy": 0.8161025867462158,
+      "eval_math_num_tokens": 243198176.0,
+      "eval_math_runtime": 35.7973,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_physics_entropy": 0.2059074219465256,
+      "eval_physics_loss": 1.4022636413574219,
+      "eval_physics_mean_token_accuracy": 0.7963426208496094,
+      "eval_physics_num_tokens": 243198176.0,
+      "eval_physics_runtime": 41.1599,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 8500
+    },
+    {
+      "entropy": 0.05167351650306955,
+      "epoch": 13.616,
+      "grad_norm": 0.478515625,
+      "learning_rate": 3.3133333333333335e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904801592230796,
+      "num_tokens": 243483881.0,
+      "step": 8510
+    },
+    {
+      "entropy": 0.053366528800688685,
+      "epoch": 13.632,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.2911111111111116e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9902401711791754,
+      "num_tokens": 243764128.0,
+      "step": 8520
+    },
+    {
+      "entropy": 0.05087098733056337,
+      "epoch": 13.648,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904976122081279,
+      "num_tokens": 244051585.0,
+      "step": 8530
+    },
+    {
+      "entropy": 0.05087883236119524,
+      "epoch": 13.664,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.2466666666666668e-06,
+      "loss": 0.0302,
+      "mean_token_accuracy": 0.9905655104666948,
+      "num_tokens": 244349004.0,
+      "step": 8540
+    },
+    {
+      "entropy": 0.04904007929144427,
+      "epoch": 13.68,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.2244444444444444e-06,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.990977830812335,
+      "num_tokens": 244649637.0,
+      "step": 8550
+    },
+    {
+      "entropy": 0.0504306229762733,
+      "epoch": 13.696,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.202222222222223e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907431546598673,
+      "num_tokens": 244940362.0,
+      "step": 8560
+    },
+    {
+      "entropy": 0.05084956689970568,
+      "epoch": 13.712,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904598340392112,
+      "num_tokens": 245228064.0,
+      "step": 8570
+    },
+    {
+      "entropy": 0.05026600864948705,
+      "epoch": 13.728,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.157777777777778e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9905784897506237,
+      "num_tokens": 245516918.0,
+      "step": 8580
+    },
+    {
+      "entropy": 0.050801993359345946,
+      "epoch": 13.744,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.1355555555555557e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907515291124582,
+      "num_tokens": 245807811.0,
+      "step": 8590
+    },
+    {
+      "entropy": 0.05082948401104659,
+      "epoch": 13.76,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.1133333333333337e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9906744692474604,
+      "num_tokens": 246099779.0,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_math_entropy": 0.17962482076883315,
+      "eval_math_loss": 1.3373396396636963,
+      "eval_math_mean_token_accuracy": 0.816063407421112,
+      "eval_math_num_tokens": 246099779.0,
+      "eval_math_runtime": 35.7911,
+      "eval_math_samples_per_second": 13.97,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_physics_entropy": 0.20641914105415343,
+      "eval_physics_loss": 1.4027777910232544,
+      "eval_physics_mean_token_accuracy": 0.796207398891449,
+      "eval_physics_num_tokens": 246099779.0,
+      "eval_physics_runtime": 41.482,
+      "eval_physics_samples_per_second": 12.053,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 8600
+    },
+    {
+      "entropy": 0.05176880229264498,
+      "epoch": 13.776,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0911111111111114e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904642947018146,
+      "num_tokens": 246383665.0,
+      "step": 8610
+    },
+    {
+      "entropy": 0.05162258397322148,
+      "epoch": 13.792,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.068888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904142130166292,
+      "num_tokens": 246668395.0,
+      "step": 8620
+    },
+    {
+      "entropy": 0.0508142440347001,
+      "epoch": 13.808,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.0466666666666666e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907273758202791,
+      "num_tokens": 246960911.0,
+      "step": 8630
+    },
+    {
+      "entropy": 0.04985311826458201,
+      "epoch": 13.824,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.024444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908174972981214,
+      "num_tokens": 247253562.0,
+      "step": 8640
+    },
+    {
+      "entropy": 0.052733641443774106,
+      "epoch": 13.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0022222222222227e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9900983639061451,
+      "num_tokens": 247529533.0,
+      "step": 8650
+    },
+    {
+      "entropy": 0.05027402397245169,
+      "epoch": 13.856,
+      "grad_norm": 0.45703125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990868479758501,
+      "num_tokens": 247819702.0,
+      "step": 8660
+    },
+    {
+      "entropy": 0.050137166888453065,
+      "epoch": 13.872,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.957777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907190252095461,
+      "num_tokens": 248113990.0,
+      "step": 8670
+    },
+    {
+      "entropy": 0.0517217613523826,
+      "epoch": 13.888,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.935555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902944087982177,
+      "num_tokens": 248397281.0,
+      "step": 8680
+    },
+    {
+      "entropy": 0.05074853647965938,
+      "epoch": 13.904,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.9133333333333335e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9904990620911122,
+      "num_tokens": 248689496.0,
+      "step": 8690
+    },
+    {
+      "entropy": 0.05088953715749085,
+      "epoch": 13.92,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.891111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905771657824516,
+      "num_tokens": 248975741.0,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_math_entropy": 0.17970398700237275,
+      "eval_math_loss": 1.336698293685913,
+      "eval_math_mean_token_accuracy": 0.8159835457801818,
+      "eval_math_num_tokens": 248975741.0,
+      "eval_math_runtime": 35.7809,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_physics_entropy": 0.20640488415956498,
+      "eval_physics_loss": 1.4024730920791626,
+      "eval_physics_mean_token_accuracy": 0.7963580441474915,
+      "eval_physics_num_tokens": 248975741.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8700
+    },
+    {
+      "entropy": 0.04994275859789923,
+      "epoch": 13.936,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.868888888888889e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9907119512557984,
+      "num_tokens": 249270221.0,
+      "step": 8710
+    },
+    {
+      "entropy": 0.04963390504708513,
+      "epoch": 13.952,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.8466666666666672e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9907703123986721,
+      "num_tokens": 249563652.0,
+      "step": 8720
+    },
+    {
+      "entropy": 0.05193966852966696,
+      "epoch": 13.968,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.824444444444445e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902710720896721,
+      "num_tokens": 249847652.0,
+      "step": 8730
+    },
+    {
+      "entropy": 0.05370221418561414,
+      "epoch": 13.984,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.8022222222222225e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9897177372127771,
+      "num_tokens": 250120462.0,
+      "step": 8740
+    },
+    {
+      "entropy": 0.05114520986098796,
+      "epoch": 14.0,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9904374986886978,
+      "num_tokens": 250409936.0,
+      "step": 8750
+    },
+    {
+      "entropy": 0.05091056008823216,
+      "epoch": 14.016,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.757777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908093482255935,
+      "num_tokens": 250696959.0,
+      "step": 8760
+    },
+    {
+      "entropy": 0.052239940082654354,
+      "epoch": 14.032,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.7355555555555557e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906717732548713,
+      "num_tokens": 250980823.0,
+      "step": 8770
+    },
+    {
+      "entropy": 0.051255598349962385,
+      "epoch": 14.048,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.7133333333333333e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990817604586482,
+      "num_tokens": 251268212.0,
+      "step": 8780
+    },
+    {
+      "entropy": 0.05210095479851588,
+      "epoch": 14.064,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.6911111111111114e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904542353004218,
+      "num_tokens": 251553952.0,
+      "step": 8790
+    },
+    {
+      "entropy": 0.05244319257326424,
+      "epoch": 14.08,
+      "grad_norm": 0.458984375,
+      "learning_rate": 2.6688888888888894e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904410917311907,
+      "num_tokens": 251838168.0,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_math_entropy": 0.17896863031387328,
+      "eval_math_loss": 1.3373428583145142,
+      "eval_math_mean_token_accuracy": 0.8160420455932618,
+      "eval_math_num_tokens": 251838168.0,
+      "eval_math_runtime": 35.776,
+      "eval_math_samples_per_second": 13.976,
+      "eval_math_steps_per_second": 3.494,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_physics_entropy": 0.20583795410394667,
+      "eval_physics_loss": 1.4028042554855347,
+      "eval_physics_mean_token_accuracy": 0.7963007631301879,
+      "eval_physics_num_tokens": 251838168.0,
+      "eval_physics_runtime": 41.4723,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8800
+    },
+    {
+      "entropy": 0.05141720774117857,
+      "epoch": 14.096,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.646666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906542908400298,
+      "num_tokens": 252121559.0,
+      "step": 8810
+    },
+    {
+      "entropy": 0.050573783030267806,
+      "epoch": 14.112,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.6244444444444446e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9908661898225546,
+      "num_tokens": 252413324.0,
+      "step": 8820
+    },
+    {
+      "entropy": 0.05072221851442009,
+      "epoch": 14.128,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.6022222222222227e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908107925206423,
+      "num_tokens": 252700652.0,
+      "step": 8830
+    },
+    {
+      "entropy": 0.04975284937536344,
+      "epoch": 14.144,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910169199109078,
+      "num_tokens": 252992593.0,
+      "step": 8840
+    },
+    {
+      "entropy": 0.049413899821229276,
+      "epoch": 14.16,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.557777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910381846129894,
+      "num_tokens": 253284586.0,
+      "step": 8850
+    },
+    {
+      "entropy": 0.04987312061712146,
+      "epoch": 14.176,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.5355555555555555e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.9910611238330602,
+      "num_tokens": 253580393.0,
+      "step": 8860
+    },
+    {
+      "entropy": 0.05002628688234836,
+      "epoch": 14.192,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.5133333333333336e-06,
+      "loss": 0.0303,
+      "mean_token_accuracy": 0.9909944098442793,
+      "num_tokens": 253873719.0,
+      "step": 8870
+    },
+    {
+      "entropy": 0.051363390649203214,
+      "epoch": 14.208,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.491111111111111e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.990709126740694,
+      "num_tokens": 254163170.0,
+      "step": 8880
+    },
+    {
+      "entropy": 0.05138509668176994,
+      "epoch": 14.224,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905464217066765,
+      "num_tokens": 254445476.0,
+      "step": 8890
+    },
+    {
+      "entropy": 0.051887833909131584,
+      "epoch": 14.24,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.446666666666667e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906827192753553,
+      "num_tokens": 254731054.0,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_math_entropy": 0.17892871648073197,
+      "eval_math_loss": 1.3377963304519653,
+      "eval_math_mean_token_accuracy": 0.8161516799926758,
+      "eval_math_num_tokens": 254731054.0,
+      "eval_math_runtime": 35.7685,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_physics_entropy": 0.20582070446014405,
+      "eval_physics_loss": 1.4032886028289795,
+      "eval_physics_mean_token_accuracy": 0.7961821508407593,
+      "eval_physics_num_tokens": 254731054.0,
+      "eval_physics_runtime": 41.4536,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8900
+    },
+    {
+      "entropy": 0.05053608596790582,
+      "epoch": 14.256,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.4244444444444444e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908701453357935,
+      "num_tokens": 255017263.0,
+      "step": 8910
+    },
+    {
+      "entropy": 0.04994903282495215,
+      "epoch": 14.272,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.4022222222222225e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907931432127952,
+      "num_tokens": 255310045.0,
+      "step": 8920
+    },
+    {
+      "entropy": 0.0498261819826439,
+      "epoch": 14.288,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.38e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908365543931723,
+      "num_tokens": 255601380.0,
+      "step": 8930
+    },
+    {
+      "entropy": 0.05181859011063352,
+      "epoch": 14.304,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.357777777777778e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906132340431213,
+      "num_tokens": 255886699.0,
+      "step": 8940
+    },
+    {
+      "entropy": 0.0507721847621724,
+      "epoch": 14.32,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.3355555555555557e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9908549074083567,
+      "num_tokens": 256174927.0,
+      "step": 8950
+    },
+    {
+      "entropy": 0.05199292111210525,
+      "epoch": 14.336,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.3133333333333333e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990635683387518,
+      "num_tokens": 256456044.0,
+      "step": 8960
+    },
+    {
+      "entropy": 0.051074851502198725,
+      "epoch": 14.352,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.2911111111111114e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907376989722252,
+      "num_tokens": 256737497.0,
+      "step": 8970
+    },
+    {
+      "entropy": 0.050048362242523584,
+      "epoch": 14.368,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.268888888888889e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909071121364832,
+      "num_tokens": 257032721.0,
+      "step": 8980
+    },
+    {
+      "entropy": 0.052360807289369406,
+      "epoch": 14.384,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.2466666666666666e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905145078897476,
+      "num_tokens": 257311728.0,
+      "step": 8990
+    },
+    {
+      "entropy": 0.051360888360068205,
+      "epoch": 14.4,
+      "grad_norm": 0.4921875,
+      "learning_rate": 2.2244444444444447e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906007140874863,
+      "num_tokens": 257598724.0,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_math_entropy": 0.17880775159597398,
+      "eval_math_loss": 1.3379243612289429,
+      "eval_math_mean_token_accuracy": 0.8160511145591736,
+      "eval_math_num_tokens": 257598724.0,
+      "eval_math_runtime": 35.7772,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_physics_entropy": 0.20577423387765884,
+      "eval_physics_loss": 1.4032671451568604,
+      "eval_physics_mean_token_accuracy": 0.7963672475814819,
+      "eval_physics_num_tokens": 257598724.0,
+      "eval_physics_runtime": 41.5292,
+      "eval_physics_samples_per_second": 12.04,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 9000
+    },
+    {
+      "entropy": 0.05165425557643175,
+      "epoch": 14.416,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.2022222222222227e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904982160776854,
+      "num_tokens": 257881999.0,
+      "step": 9010
+    },
+    {
+      "entropy": 0.05141539016040042,
+      "epoch": 14.432,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906045537441969,
+      "num_tokens": 258168215.0,
+      "step": 9020
+    },
+    {
+      "entropy": 0.050296508916653694,
+      "epoch": 14.448,
+      "grad_norm": 0.375,
+      "learning_rate": 2.157777777777778e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907093074172735,
+      "num_tokens": 258455416.0,
+      "step": 9030
+    },
+    {
+      "entropy": 0.050917042326182126,
+      "epoch": 14.464,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.1355555555555555e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907846868038177,
+      "num_tokens": 258743229.0,
+      "step": 9040
+    },
+    {
+      "entropy": 0.050571117317304015,
+      "epoch": 14.48,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.1133333333333336e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907037664204836,
+      "num_tokens": 259028366.0,
+      "step": 9050
+    },
+    {
+      "entropy": 0.05255825007334351,
+      "epoch": 14.496,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.091111111111111e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990223740413785,
+      "num_tokens": 259304539.0,
+      "step": 9060
+    },
+    {
+      "entropy": 0.04984008674509823,
+      "epoch": 14.512,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.0688888888888892e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909922797232866,
+      "num_tokens": 259596545.0,
+      "step": 9070
+    },
+    {
+      "entropy": 0.05399213619530201,
+      "epoch": 14.528,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.046666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901899788528681,
+      "num_tokens": 259870582.0,
+      "step": 9080
+    },
+    {
+      "entropy": 0.05196048454381526,
+      "epoch": 14.544,
+      "grad_norm": 0.47265625,
+      "learning_rate": 2.024444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904056012630462,
+      "num_tokens": 260152920.0,
+      "step": 9090
+    },
+    {
+      "entropy": 0.05186033591162413,
+      "epoch": 14.56,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.0022222222222225e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905133057385683,
+      "num_tokens": 260437753.0,
+      "step": 9100
+    },
+    {
+      "epoch": 14.56,
+      "eval_math_entropy": 0.1789300767183304,
+      "eval_math_loss": 1.337916612625122,
+      "eval_math_mean_token_accuracy": 0.8160979781150818,
+      "eval_math_num_tokens": 260437753.0,
+      "eval_math_runtime": 35.8923,
+      "eval_math_samples_per_second": 13.931,
+      "eval_math_steps_per_second": 3.483,
+      "step": 9100
+    },
+    {
+      "epoch": 14.56,
+      "eval_physics_entropy": 0.2058382231593132,
+      "eval_physics_loss": 1.4032976627349854,
+      "eval_physics_mean_token_accuracy": 0.796412097454071,
+      "eval_physics_num_tokens": 260437753.0,
+      "eval_physics_runtime": 41.6691,
+      "eval_physics_samples_per_second": 11.999,
+      "eval_physics_steps_per_second": 3.0,
+      "step": 9100
+    },
+    {
+      "entropy": 0.05178943979553878,
+      "epoch": 14.576,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.98e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905171319842339,
+      "num_tokens": 260719119.0,
+      "step": 9110
+    },
+    {
+      "entropy": 0.05121098265517503,
+      "epoch": 14.592,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.9577777777777777e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907041292637586,
+      "num_tokens": 261005643.0,
+      "step": 9120
+    },
+    {
+      "entropy": 0.05071252475026995,
+      "epoch": 14.608,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.9355555555555558e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908163469284773,
+      "num_tokens": 261297751.0,
+      "step": 9130
+    },
+    {
+      "entropy": 0.052167282067239286,
+      "epoch": 14.624,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.9133333333333334e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905383501201868,
+      "num_tokens": 261574524.0,
+      "step": 9140
+    },
+    {
+      "entropy": 0.05092277757357806,
+      "epoch": 14.64,
+      "grad_norm": 0.4609375,
+      "learning_rate": 1.8911111111111114e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906485483050347,
+      "num_tokens": 261860827.0,
+      "step": 9150
+    },
+    {
+      "entropy": 0.0508843480842188,
+      "epoch": 14.656,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.868888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905889291316271,
+      "num_tokens": 262146577.0,
+      "step": 9160
+    },
+    {
+      "entropy": 0.051762618916109206,
+      "epoch": 14.672,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.8466666666666668e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990484931319952,
+      "num_tokens": 262425056.0,
+      "step": 9170
+    },
+    {
+      "entropy": 0.050335659622214736,
+      "epoch": 14.688,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.8244444444444445e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908275879919529,
+      "num_tokens": 262718069.0,
+      "step": 9180
+    },
+    {
+      "entropy": 0.05197438263567165,
+      "epoch": 14.704,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.8022222222222225e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990571141988039,
+      "num_tokens": 263005354.0,
+      "step": 9190
+    },
+    {
+      "entropy": 0.050634497124701736,
+      "epoch": 14.72,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9907538581639528,
+      "num_tokens": 263290156.0,
+      "step": 9200
+    },
+    {
+      "epoch": 14.72,
+      "eval_math_entropy": 0.17962117153406143,
+      "eval_math_loss": 1.3382362127304077,
+      "eval_math_mean_token_accuracy": 0.8159899268150329,
+      "eval_math_num_tokens": 263290156.0,
+      "eval_math_runtime": 35.8309,
+      "eval_math_samples_per_second": 13.954,
+      "eval_math_steps_per_second": 3.489,
+      "step": 9200
+    },
+    {
+      "epoch": 14.72,
+      "eval_physics_entropy": 0.20646389347314834,
+      "eval_physics_loss": 1.4038169384002686,
+      "eval_physics_mean_token_accuracy": 0.7962654252052307,
+      "eval_physics_num_tokens": 263290156.0,
+      "eval_physics_runtime": 41.5336,
+      "eval_physics_samples_per_second": 12.038,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 9200
+    },
+    {
+      "entropy": 0.05221541177015752,
+      "epoch": 14.736,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.757777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905163571238518,
+      "num_tokens": 263570163.0,
+      "step": 9210
+    },
+    {
+      "entropy": 0.05265975071815774,
+      "epoch": 14.752,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.7355555555555555e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901549957692624,
+      "num_tokens": 263847082.0,
+      "step": 9220
+    },
+    {
+      "entropy": 0.05067356282379478,
+      "epoch": 14.768,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.7133333333333336e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906387656927109,
+      "num_tokens": 264135491.0,
+      "step": 9230
+    },
+    {
+      "entropy": 0.05214657548349351,
+      "epoch": 14.784,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.6911111111111112e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9905820898711681,
+      "num_tokens": 264423201.0,
+      "step": 9240
+    },
+    {
+      "entropy": 0.05079062914010137,
+      "epoch": 14.8,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.668888888888889e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907969791442156,
+      "num_tokens": 264711872.0,
+      "step": 9250
+    },
+    {
+      "entropy": 0.05266349882585928,
+      "epoch": 14.816,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.6466666666666666e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903959967195988,
+      "num_tokens": 264991504.0,
+      "step": 9260
+    },
+    {
+      "entropy": 0.05068820312153548,
+      "epoch": 14.832,
+      "grad_norm": 0.375,
+      "learning_rate": 1.6244444444444447e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907281890511512,
+      "num_tokens": 265283873.0,
+      "step": 9270
+    },
+    {
+      "entropy": 0.05143596041016281,
+      "epoch": 14.848,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.6022222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905165329575538,
+      "num_tokens": 265563990.0,
+      "step": 9280
+    },
+    {
+      "entropy": 0.05017106860177591,
+      "epoch": 14.864,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906800303608179,
+      "num_tokens": 265850179.0,
+      "step": 9290
+    },
+    {
+      "entropy": 0.04978077416308224,
+      "epoch": 14.88,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.5577777777777777e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.9907668430358172,
+      "num_tokens": 266145551.0,
+      "step": 9300
+    },
+    {
+      "epoch": 14.88,
+      "eval_math_entropy": 0.17940663027763368,
+      "eval_math_loss": 1.3379700183868408,
+      "eval_math_mean_token_accuracy": 0.8160032463073731,
+      "eval_math_num_tokens": 266145551.0,
+      "eval_math_runtime": 35.8577,
+      "eval_math_samples_per_second": 13.944,
+      "eval_math_steps_per_second": 3.486,
+      "step": 9300
+    },
+    {
+      "epoch": 14.88,
+      "eval_physics_entropy": 0.20622114306688308,
+      "eval_physics_loss": 1.4035049676895142,
+      "eval_physics_mean_token_accuracy": 0.7961468186378479,
+      "eval_physics_num_tokens": 266145551.0,
+      "eval_physics_runtime": 41.5608,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 9300
+    },
+    {
+      "entropy": 0.050715955626219514,
+      "epoch": 14.896,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.5355555555555558e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907834134995938,
+      "num_tokens": 266432255.0,
+      "step": 9310
+    },
+    {
+      "entropy": 0.0511104385368526,
+      "epoch": 14.912,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.5133333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906729698181153,
+      "num_tokens": 266719617.0,
+      "step": 9320
+    },
+    {
+      "entropy": 0.0505100853391923,
+      "epoch": 14.928,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.4911111111111112e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907874137163162,
+      "num_tokens": 267012400.0,
+      "step": 9330
+    },
+    {
+      "entropy": 0.05214700745418668,
+      "epoch": 14.943999999999999,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.468888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904575120657683,
+      "num_tokens": 267295468.0,
+      "step": 9340
+    },
+    {
+      "entropy": 0.051390086626634,
+      "epoch": 14.96,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.4466666666666669e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905224211513997,
+      "num_tokens": 267580054.0,
+      "step": 9350
+    },
+    {
+      "entropy": 0.050726299383677544,
+      "epoch": 14.975999999999999,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.4244444444444447e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9908859673887491,
+      "num_tokens": 267871906.0,
+      "step": 9360
+    },
+    {
+      "entropy": 0.05101369803305715,
+      "epoch": 14.992,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.4022222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905830901116133,
+      "num_tokens": 268156343.0,
+      "step": 9370
+    },
+    {
+      "entropy": 0.051476930570788684,
+      "epoch": 15.008,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904872190207243,
+      "num_tokens": 268442263.0,
+      "step": 9380
+    },
+    {
+      "entropy": 0.05267978028859943,
+      "epoch": 15.024,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.357777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990261047706008,
+      "num_tokens": 268714841.0,
+      "step": 9390
+    },
+    {
+      "entropy": 0.05263522565364838,
+      "epoch": 15.04,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.3355555555555558e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904483895748853,
+      "num_tokens": 268994210.0,
+      "step": 9400
+    },
+    {
+      "epoch": 15.04,
+      "eval_math_entropy": 0.17955057173967362,
+      "eval_math_loss": 1.3379791975021362,
+      "eval_math_mean_token_accuracy": 0.8160371327400208,
+      "eval_math_num_tokens": 268994210.0,
+      "eval_math_runtime": 35.8381,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 9400
+    },
+    {
+      "epoch": 15.04,
+      "eval_physics_entropy": 0.20631873172521592,
+      "eval_physics_loss": 1.403602957725525,
+      "eval_physics_mean_token_accuracy": 0.7963659038543701,
+      "eval_physics_num_tokens": 268994210.0,
+      "eval_physics_runtime": 41.5687,
+      "eval_physics_samples_per_second": 12.028,
+      "eval_physics_steps_per_second": 3.007,
+      "step": 9400
+    },
+    {
+      "entropy": 0.05078244159230962,
+      "epoch": 15.056,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.3133333333333334e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910456586629153,
+      "num_tokens": 269285822.0,
+      "step": 9410
+    },
+    {
+      "entropy": 0.04946443063672632,
+      "epoch": 15.072,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.2911111111111112e-06,
+      "loss": 0.0303,
+      "mean_token_accuracy": 0.9911473792046308,
+      "num_tokens": 269579649.0,
+      "step": 9420
+    },
+    {
+      "entropy": 0.050997486570850015,
+      "epoch": 15.088,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.268888888888889e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908084504306316,
+      "num_tokens": 269866946.0,
+      "step": 9430
+    },
+    {
+      "entropy": 0.051223152736201885,
+      "epoch": 15.104,
+      "grad_norm": 0.484375,
+      "learning_rate": 1.2466666666666667e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909345023334026,
+      "num_tokens": 270151076.0,
+      "step": 9440
+    },
+    {
+      "entropy": 0.05211627440294251,
+      "epoch": 15.12,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.2244444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906228743493557,
+      "num_tokens": 270433447.0,
+      "step": 9450
+    },
+    {
+      "entropy": 0.05138029637746513,
+      "epoch": 15.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.2022222222222223e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908742789179087,
+      "num_tokens": 270721393.0,
+      "step": 9460
+    },
+    {
+      "entropy": 0.05007477108156309,
+      "epoch": 15.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.9911369781941175,
+      "num_tokens": 271018581.0,
+      "step": 9470
+    },
+    {
+      "entropy": 0.050468166405335066,
+      "epoch": 15.168,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.1577777777777778e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908524744212628,
+      "num_tokens": 271307471.0,
+      "step": 9480
+    },
+    {
+      "entropy": 0.05269346619024873,
+      "epoch": 15.184,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.1355555555555558e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904394499957562,
+      "num_tokens": 271590009.0,
+      "step": 9490
+    },
+    {
+      "entropy": 0.050732248765416446,
+      "epoch": 15.2,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.1133333333333334e-06,
+      "loss": 0.0302,
+      "mean_token_accuracy": 0.9909780859947205,
+      "num_tokens": 271882895.0,
+      "step": 9500
+    },
+    {
+      "epoch": 15.2,
+      "eval_math_entropy": 0.17890966862440108,
+      "eval_math_loss": 1.3383686542510986,
+      "eval_math_mean_token_accuracy": 0.8160392146110534,
+      "eval_math_num_tokens": 271882895.0,
+      "eval_math_runtime": 35.8479,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 9500
+    },
+    {
+      "epoch": 15.2,
+      "eval_physics_entropy": 0.205828775703907,
+      "eval_physics_loss": 1.4038267135620117,
+      "eval_physics_mean_token_accuracy": 0.7963502130508423,
+      "eval_physics_num_tokens": 271882895.0,
+      "eval_physics_runtime": 41.1707,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 9500
+    },
+    {
+      "entropy": 0.05281253398861736,
+      "epoch": 15.216,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0911111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902109578251839,
+      "num_tokens": 272155113.0,
+      "step": 9510
+    },
+    {
+      "entropy": 0.05095949907554313,
+      "epoch": 15.232,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.068888888888889e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9907589819282293,
+      "num_tokens": 272437443.0,
+      "step": 9520
+    },
+    {
+      "entropy": 0.05068987334379926,
+      "epoch": 15.248,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.0466666666666669e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9911098714917899,
+      "num_tokens": 272727519.0,
+      "step": 9530
+    },
+    {
+      "entropy": 0.05247048016171903,
+      "epoch": 15.264,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.0244444444444445e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904446382075548,
+      "num_tokens": 273009301.0,
+      "step": 9540
+    },
+    {
+      "entropy": 0.05246194021310657,
+      "epoch": 15.28,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.0022222222222223e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904823791235685,
+      "num_tokens": 273290197.0,
+      "step": 9550
+    },
+    {
+      "entropy": 0.05017303994391113,
+      "epoch": 15.296,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909063238650561,
+      "num_tokens": 273578881.0,
+      "step": 9560
+    },
+    {
+      "entropy": 0.05100858648074791,
+      "epoch": 15.312,
+      "grad_norm": 0.419921875,
+      "learning_rate": 9.57777777777778e-07,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9906195200979709,
+      "num_tokens": 273868038.0,
+      "step": 9570
+    },
+    {
+      "entropy": 0.05174001138657332,
+      "epoch": 15.328,
+      "grad_norm": 0.427734375,
+      "learning_rate": 9.355555555555557e-07,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.990441370382905,
+      "num_tokens": 274149762.0,
+      "step": 9580
+    },
+    {
+      "entropy": 0.051202584989368916,
+      "epoch": 15.344,
+      "grad_norm": 0.451171875,
+      "learning_rate": 9.133333333333334e-07,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906471762806177,
+      "num_tokens": 274432517.0,
+      "step": 9590
+    },
+    {
+      "entropy": 0.049917721375823024,
+      "epoch": 15.36,
+      "grad_norm": 0.4375,
+      "learning_rate": 8.911111111111112e-07,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.99101542532444,
+      "num_tokens": 274726803.0,
+      "step": 9600
+    },
+    {
+      "epoch": 15.36,
+      "eval_math_entropy": 0.1788821375966072,
+      "eval_math_loss": 1.338241457939148,
+      "eval_math_mean_token_accuracy": 0.816077757358551,
+      "eval_math_num_tokens": 274726803.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 9600
+    },
+    {
+      "epoch": 15.36,
+      "eval_physics_entropy": 0.20582045251131056,
+      "eval_physics_loss": 1.4037485122680664,
+      "eval_physics_mean_token_accuracy": 0.796428108215332,
+      "eval_physics_num_tokens": 274726803.0,
+      "eval_physics_runtime": 41.502,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 9600
+    },
+    {
+      "entropy": 0.05084420018829405,
+      "epoch": 15.376,
+      "grad_norm": 0.4765625,
+      "learning_rate": 8.68888888888889e-07,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909689243882894,
+      "num_tokens": 275017378.0,
+      "step": 9610
+    },
+    {
+      "entropy": 0.05047262012958527,
+      "epoch": 15.392,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.466666666666668e-07,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.990967021510005,
+      "num_tokens": 275306836.0,
+      "step": 9620
+    },
+    {
+      "entropy": 0.051559871272183956,
+      "epoch": 15.408,
+      "grad_norm": 0.42578125,
+      "learning_rate": 8.244444444444445e-07,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906136877834797,
+      "num_tokens": 275593244.0,
+      "step": 9630
+    },
+    {
+      "entropy": 0.05223148600198328,
+      "epoch": 15.424,
+      "grad_norm": 0.462890625,
+      "learning_rate": 8.022222222222223e-07,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905043765902519,
+      "num_tokens": 275871246.0,
+      "step": 9640
+    },
+    {
+      "entropy": 0.05090933956671506,
+      "epoch": 15.44,
+      "grad_norm": 0.4453125,
+      "learning_rate": 7.8e-07,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906908594071865,
+      "num_tokens": 276155963.0,
+      "step": 9650
+    },
+    {
+      "entropy": 0.05163709740154445,
+      "epoch": 15.456,
+      "grad_norm": 0.478515625,
+      "learning_rate": 7.577777777777779e-07,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906673815101386,
+      "num_tokens": 276437404.0,
+      "step": 9660
+    },
+    {
+      "entropy": 0.050778869073837996,
+      "epoch": 15.472,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.355555555555556e-07,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9908820949494839,
+      "num_tokens": 276725554.0,
+      "step": 9670
+    },
+    {
+      "entropy": 0.0514642862835899,
+      "epoch": 15.488,
+      "grad_norm": 0.474609375,
+      "learning_rate": 7.133333333333334e-07,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9905849743634463,
+      "num_tokens": 277010524.0,
+      "step": 9680
+    },
+    {
+      "entropy": 0.050047663040459156,
+      "epoch": 15.504,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.911111111111111e-07,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990903789922595,
+      "num_tokens": 277301488.0,
+      "step": 9690
+    },
+    {
+      "entropy": 0.0512123049935326,
+      "epoch": 15.52,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.68888888888889e-07,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906671985983848,
+      "num_tokens": 277592951.0,
+      "step": 9700
+    },
+    {
+      "epoch": 15.52,
+      "eval_math_entropy": 0.17894334828853606,
+      "eval_math_loss": 1.3383890390396118,
+      "eval_math_mean_token_accuracy": 0.8160645146369934,
+      "eval_math_num_tokens": 277592951.0,
+      "eval_math_runtime": 35.7927,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 9700
+    },
+    {
+      "epoch": 15.52,
+      "eval_physics_entropy": 0.20584783589839936,
+      "eval_physics_loss": 1.4039112329483032,
+      "eval_physics_mean_token_accuracy": 0.796358169078827,
+      "eval_physics_num_tokens": 277592951.0,
+      "eval_physics_runtime": 41.4831,
+      "eval_physics_samples_per_second": 12.053,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 9700
+    },
+    {
+      "entropy": 0.048784788453485814,
+      "epoch": 15.536,
+      "grad_norm": 0.375,
+      "learning_rate": 6.466666666666667e-07,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.9912266310304403,
+      "num_tokens": 277892234.0,
+      "step": 9710
+    },
+    {
+      "entropy": 0.0524211410083808,
+      "epoch": 15.552,
+      "grad_norm": 0.43359375,
+      "learning_rate": 6.244444444444445e-07,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903773467987775,
+      "num_tokens": 278171426.0,
+      "step": 9720
+    },
+    {
+      "entropy": 0.051696876191999766,
+      "epoch": 15.568,
+      "grad_norm": 0.462890625,
+      "learning_rate": 6.022222222222223e-07,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9907380498945713,
+      "num_tokens": 278455973.0,
+      "step": 9730
+    },
+    {
+      "entropy": 0.05162487926427275,
+      "epoch": 15.584,
+      "grad_norm": 0.4296875,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904108371585607,
+      "num_tokens": 278736136.0,
+      "step": 9740
+    },
+    {
+      "entropy": 0.050188562797848134,
+      "epoch": 15.6,
+      "grad_norm": 0.470703125,
+      "learning_rate": 5.577777777777779e-07,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.99085229113698,
+      "num_tokens": 279025021.0,
+      "step": 9750
+    },
+    {
+      "entropy": 0.0510073521756567,
+      "epoch": 15.616,
+      "grad_norm": 0.498046875,
+      "learning_rate": 5.355555555555556e-07,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990767364948988,
+      "num_tokens": 279315812.0,
+      "step": 9760
+    },
+    {
+      "entropy": 0.05096343404147774,
+      "epoch": 15.632,
+      "grad_norm": 0.44140625,
+      "learning_rate": 5.133333333333334e-07,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907372202724218,
+      "num_tokens": 279602382.0,
+      "step": 9770
+    },
+    {
+      "entropy": 0.051482615130953494,
+      "epoch": 15.648,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.911111111111112e-07,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905079193413258,
+      "num_tokens": 279888105.0,
+      "step": 9780
+    },
+    {
+      "entropy": 0.05247906430158764,
+      "epoch": 15.664,
+      "grad_norm": 0.427734375,
+      "learning_rate": 4.688888888888889e-07,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905213572084903,
+      "num_tokens": 280166994.0,
+      "step": 9790
+    },
+    {
+      "entropy": 0.05217490267241374,
+      "epoch": 15.68,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.466666666666667e-07,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905618295073509,
+      "num_tokens": 280452818.0,
+      "step": 9800
+    },
+    {
+      "epoch": 15.68,
+      "eval_math_entropy": 0.1789257292151451,
+      "eval_math_loss": 1.3384068012237549,
+      "eval_math_mean_token_accuracy": 0.8160247092247009,
+      "eval_math_num_tokens": 280452818.0,
+      "eval_math_runtime": 35.9093,
+      "eval_math_samples_per_second": 13.924,
+      "eval_math_steps_per_second": 3.481,
+      "step": 9800
+    },
+    {
+      "epoch": 15.68,
+      "eval_physics_entropy": 0.20582703042030334,
+      "eval_physics_loss": 1.4039790630340576,
+      "eval_physics_mean_token_accuracy": 0.7964057297706604,
+      "eval_physics_num_tokens": 280452818.0,
+      "eval_physics_runtime": 41.5752,
+      "eval_physics_samples_per_second": 12.026,
+      "eval_physics_steps_per_second": 3.007,
+      "step": 9800
+    },
+    {
+      "entropy": 0.05222263811156154,
+      "epoch": 15.696,
+      "grad_norm": 0.4921875,
+      "learning_rate": 4.2444444444444447e-07,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906012628227472,
+      "num_tokens": 280732279.0,
+      "step": 9810
+    },
+    {
+      "entropy": 0.05261244579451159,
+      "epoch": 15.712,
+      "grad_norm": 0.478515625,
+      "learning_rate": 4.0222222222222224e-07,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9906971897929907,
+      "num_tokens": 281012196.0,
+      "step": 9820
+    },
+    {
+      "entropy": 0.051492129743564874,
+      "epoch": 15.728,
+      "grad_norm": 0.478515625,
+      "learning_rate": 3.8e-07,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9906229086220264,
+      "num_tokens": 281301358.0,
+      "step": 9830
+    },
+    {
+      "entropy": 0.049157703714445236,
+      "epoch": 15.744,
+      "grad_norm": 0.453125,
+      "learning_rate": 3.577777777777778e-07,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9912415307015181,
+      "num_tokens": 281597592.0,
+      "step": 9840
+    },
+    {
+      "entropy": 0.050756015651859346,
+      "epoch": 15.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.3555555555555556e-07,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990971352532506,
+      "num_tokens": 281885453.0,
+      "step": 9850
+    },
+    {
+      "entropy": 0.049710268969647586,
+      "epoch": 15.776,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.1333333333333333e-07,
+      "loss": 0.0299,
+      "mean_token_accuracy": 0.9911037530750036,
+      "num_tokens": 282182110.0,
+      "step": 9860
+    },
+    {
+      "entropy": 0.05201074434444308,
+      "epoch": 15.792,
+      "grad_norm": 0.44140625,
+      "learning_rate": 2.9111111111111116e-07,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906283039599657,
+      "num_tokens": 282467359.0,
+      "step": 9870
+    },
+    {
+      "entropy": 0.05186814813641831,
+      "epoch": 15.808,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.6888888888888893e-07,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906272884458304,
+      "num_tokens": 282753275.0,
+      "step": 9880
+    },
+    {
+      "entropy": 0.05055844767484814,
+      "epoch": 15.824,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.466666666666667e-07,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907678700983524,
+      "num_tokens": 283040141.0,
+      "step": 9890
+    },
+    {
+      "entropy": 0.05105671789497137,
+      "epoch": 15.84,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.2444444444444445e-07,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907285198569298,
+      "num_tokens": 283322369.0,
+      "step": 9900
+    },
+    {
+      "epoch": 15.84,
+      "eval_math_entropy": 0.17960893189907073,
+      "eval_math_loss": 1.3384406566619873,
+      "eval_math_mean_token_accuracy": 0.8160690641403199,
+      "eval_math_num_tokens": 283322369.0,
+      "eval_math_runtime": 35.8232,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 9900
+    },
+    {
+      "epoch": 15.84,
+      "eval_physics_entropy": 0.20642443406581878,
+      "eval_physics_loss": 1.4037851095199585,
+      "eval_physics_mean_token_accuracy": 0.7964150753021241,
+      "eval_physics_num_tokens": 283322369.0,
+      "eval_physics_runtime": 41.4999,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 9900
+    },
+    {
+      "entropy": 0.050554721464868634,
+      "epoch": 15.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.0222222222222222e-07,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907786782830954,
+      "num_tokens": 283610795.0,
+      "step": 9910
+    },
+    {
+      "entropy": 0.051272003818303345,
+      "epoch": 15.872,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.8e-07,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907614074647426,
+      "num_tokens": 283896874.0,
+      "step": 9920
+    },
+    {
+      "entropy": 0.050100649206433444,
+      "epoch": 15.888,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.577777777777778e-07,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909307535737752,
+      "num_tokens": 284192606.0,
+      "step": 9930
+    },
+    {
+      "entropy": 0.05114874708233401,
+      "epoch": 15.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.3555555555555557e-07,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907328754663467,
+      "num_tokens": 284475029.0,
+      "step": 9940
+    },
+    {
+      "entropy": 0.0502239034161903,
+      "epoch": 15.92,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.1333333333333336e-07,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908566799014806,
+      "num_tokens": 284765326.0,
+      "step": 9950
+    },
+    {
+      "entropy": 0.05247808927670121,
+      "epoch": 15.936,
+      "grad_norm": 0.419921875,
+      "learning_rate": 9.111111111111113e-08,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904217857867479,
+      "num_tokens": 285044463.0,
+      "step": 9960
+    },
+    {
+      "entropy": 0.052579887211322784,
+      "epoch": 15.952,
+      "grad_norm": 0.396484375,
+      "learning_rate": 6.888888888888889e-08,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906398896127939,
+      "num_tokens": 285324490.0,
+      "step": 9970
+    },
+    {
+      "entropy": 0.052448914595879616,
+      "epoch": 15.968,
+      "grad_norm": 0.462890625,
+      "learning_rate": 4.6666666666666674e-08,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906637586653233,
+      "num_tokens": 285606337.0,
+      "step": 9980
+    },
+    {
+      "entropy": 0.05065399822779,
+      "epoch": 15.984,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.4444444444444447e-08,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908808015286923,
+      "num_tokens": 285893562.0,
+      "step": 9990
+    },
+    {
+      "entropy": 0.05041085070697591,
+      "epoch": 16.0,
+      "grad_norm": 0.462890625,
+      "learning_rate": 2.2222222222222225e-09,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908820543438196,
+      "num_tokens": 286182784.0,
+      "step": 10000
+    },
+    {
+      "epoch": 16.0,
+      "eval_math_entropy": 0.17890032547712326,
+      "eval_math_loss": 1.3384580612182617,
+      "eval_math_mean_token_accuracy": 0.8159227700233459,
+      "eval_math_num_tokens": 286182784.0,
+      "eval_math_runtime": 35.995,
+      "eval_math_samples_per_second": 13.891,
+      "eval_math_steps_per_second": 3.473,
+      "step": 10000
+    },
+    {
+      "epoch": 16.0,
+      "eval_physics_entropy": 0.20580756539106368,
+      "eval_physics_loss": 1.403875708580017,
+      "eval_physics_mean_token_accuracy": 0.7962533917427063,
+      "eval_physics_num_tokens": 286182784.0,
+      "eval_physics_runtime": 41.2248,
+      "eval_physics_samples_per_second": 12.129,
+      "eval_physics_steps_per_second": 3.032,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.1699157606765707e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-10000/training_args.bin b/checkpoint-10000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-10000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-1500/chat_template.jinja b/checkpoint-1500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-1500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-1500/config.json b/checkpoint-1500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-1500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-1500/generation_config.json b/checkpoint-1500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-1500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-1500/model-00001-of-00004.safetensors b/checkpoint-1500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-1500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-1500/model-00002-of-00004.safetensors b/checkpoint-1500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-1500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-1500/model-00003-of-00004.safetensors b/checkpoint-1500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..70f5586
--- /dev/null
+++ b/checkpoint-1500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:621ac4d6aa3e473f22d3a38812a7504138a73d019702313a4919117ea8aaec3e
+size 4962221464
diff --git a/checkpoint-1500/model-00004-of-00004.safetensors b/checkpoint-1500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..05407dd
--- /dev/null
+++ b/checkpoint-1500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34b3a6ae7c6ae6d16ee44a370362ca0a7b83cd1e49f2801723e96f41015120e5
+size 3670322200
diff --git a/checkpoint-1500/model.safetensors.index.json b/checkpoint-1500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-1500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-1500/optimizer.pt b/checkpoint-1500/optimizer.pt
new file mode 100644
index 0000000..c0b3f0d
--- /dev/null
+++ b/checkpoint-1500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9aaadc5f4bcbd0a0a140d19c5be0b7afd243c1b8fcca346d99a3de079b99c44
+size 7927918951
diff --git a/checkpoint-1500/rng_state.pth b/checkpoint-1500/rng_state.pth
new file mode 100644
index 0000000..0ef6228
--- /dev/null
+++ b/checkpoint-1500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8fbc49139607d0b86d00edfbf77b5edc685a63d41fd8c3ac00bec6dfe6ad4f39
+size 14645
diff --git a/checkpoint-1500/scheduler.pt b/checkpoint-1500/scheduler.pt
new file mode 100644
index 0000000..3394e2c
--- /dev/null
+++ b/checkpoint-1500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb8afd6405fc2f72bcdc3e89147a10597d475bf4457b8f98d9c27aa5442f3ef9
+size 1465
diff --git a/checkpoint-1500/special_tokens_map.json b/checkpoint-1500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-1500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-1500/tokenizer.json b/checkpoint-1500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-1500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-1500/tokenizer.model b/checkpoint-1500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-1500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-1500/tokenizer_config.json b/checkpoint-1500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-1500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-1500/trainer_state.json b/checkpoint-1500/trainer_state.json
new file mode 100644
index 0000000..22a26ee
--- /dev/null
+++ b/checkpoint-1500/trainer_state.json
@@ -0,0 +1,1864 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.4,
+  "eval_steps": 100,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.2608689443833037e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-1500/training_args.bin b/checkpoint-1500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-1500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-2000/chat_template.jinja b/checkpoint-2000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-2000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-2000/config.json b/checkpoint-2000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-2000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-2000/generation_config.json b/checkpoint-2000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-2000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-2000/model-00001-of-00004.safetensors b/checkpoint-2000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-2000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-2000/model-00002-of-00004.safetensors b/checkpoint-2000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-2000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-2000/model-00003-of-00004.safetensors b/checkpoint-2000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..5a4ff1d
--- /dev/null
+++ b/checkpoint-2000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cad4d5e8ecc0a2bd329fb01fc5c2bc8a91d12509151b6b86cf402b4d8fefd33b
+size 4962221464
diff --git a/checkpoint-2000/model-00004-of-00004.safetensors b/checkpoint-2000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..e0f2a78
--- /dev/null
+++ b/checkpoint-2000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0eac13088094a819d442a7e42c58985d2e9ce6d24c96c4890b8605df548b0062
+size 3670322200
diff --git a/checkpoint-2000/model.safetensors.index.json b/checkpoint-2000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-2000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-2000/optimizer.pt b/checkpoint-2000/optimizer.pt
new file mode 100644
index 0000000..ba4c4a8
--- /dev/null
+++ b/checkpoint-2000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e5dfa11aa0553bf5c8d488be1f2ee7401cdb6114779c5ab8ffab85c32698350
+size 7927918951
diff --git a/checkpoint-2000/rng_state.pth b/checkpoint-2000/rng_state.pth
new file mode 100644
index 0000000..8408ab4
--- /dev/null
+++ b/checkpoint-2000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b900edc205652b612b068d20ee1ff0a34fe55afa453cf50078fa4f7abb881633
+size 14645
diff --git a/checkpoint-2000/scheduler.pt b/checkpoint-2000/scheduler.pt
new file mode 100644
index 0000000..bec1b44
--- /dev/null
+++ b/checkpoint-2000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e95ad2f3950cee1b83b106cf88750c8604566acab211d78973b3b5c4917b1a30
+size 1465
diff --git a/checkpoint-2000/special_tokens_map.json b/checkpoint-2000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-2000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-2000/tokenizer.json b/checkpoint-2000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-2000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-2000/tokenizer.model b/checkpoint-2000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-2000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-2000/tokenizer_config.json b/checkpoint-2000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-2000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-2000/trainer_state.json b/checkpoint-2000/trainer_state.json
new file mode 100644
index 0000000..bee5974
--- /dev/null
+++ b/checkpoint-2000/trainer_state.json
@@ -0,0 +1,2474 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.2,
+  "eval_steps": 100,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.343629994565034e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-2000/training_args.bin b/checkpoint-2000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-2000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-2500/chat_template.jinja b/checkpoint-2500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-2500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-2500/config.json b/checkpoint-2500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-2500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-2500/generation_config.json b/checkpoint-2500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-2500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-2500/model-00001-of-00004.safetensors b/checkpoint-2500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-2500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-2500/model-00002-of-00004.safetensors b/checkpoint-2500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-2500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-2500/model-00003-of-00004.safetensors b/checkpoint-2500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..6028a53
--- /dev/null
+++ b/checkpoint-2500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:25a14023d12eed245e4899af3ce46813a6494d5b568690ccc65912b88a14b620
+size 4962221464
diff --git a/checkpoint-2500/model-00004-of-00004.safetensors b/checkpoint-2500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..a71cdaa
--- /dev/null
+++ b/checkpoint-2500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c45282ac1a070f7102cf486fe0fcc4820e0389873a177b963c08fa3aec6b283a
+size 3670322200
diff --git a/checkpoint-2500/model.safetensors.index.json b/checkpoint-2500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-2500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-2500/optimizer.pt b/checkpoint-2500/optimizer.pt
new file mode 100644
index 0000000..b61c928
--- /dev/null
+++ b/checkpoint-2500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:67b4eb0bbee5c0e642c3831c6df54e5445d18d09125044af3b6e791e5bdc409e
+size 7927918951
diff --git a/checkpoint-2500/rng_state.pth b/checkpoint-2500/rng_state.pth
new file mode 100644
index 0000000..c78208e
--- /dev/null
+++ b/checkpoint-2500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:add22108b7fb63b2a784432ac9ae752c1cd84fe3d5d8cd63573a474556580d37
+size 14645
diff --git a/checkpoint-2500/scheduler.pt b/checkpoint-2500/scheduler.pt
new file mode 100644
index 0000000..444872d
--- /dev/null
+++ b/checkpoint-2500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd200143f139327b4d41bcb99ffeaa13b2acc11ded5b3dbab181a6ff2ce855d7
+size 1465
diff --git a/checkpoint-2500/special_tokens_map.json b/checkpoint-2500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-2500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-2500/tokenizer.json b/checkpoint-2500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-2500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-2500/tokenizer.model b/checkpoint-2500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-2500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-2500/tokenizer_config.json b/checkpoint-2500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-2500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-2500/trainer_state.json b/checkpoint-2500/trainer_state.json
new file mode 100644
index 0000000..81c959a
--- /dev/null
+++ b/checkpoint-2500/trainer_state.json
@@ -0,0 +1,3084 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "eval_steps": 100,
+  "global_step": 2500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 5.428120947106271e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-2500/training_args.bin b/checkpoint-2500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-2500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-3000/chat_template.jinja b/checkpoint-3000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-3000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-3000/config.json b/checkpoint-3000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-3000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-3000/generation_config.json b/checkpoint-3000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-3000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-3000/model-00001-of-00004.safetensors b/checkpoint-3000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-3000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-3000/model-00002-of-00004.safetensors b/checkpoint-3000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-3000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-3000/model-00003-of-00004.safetensors b/checkpoint-3000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..9353267
--- /dev/null
+++ b/checkpoint-3000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4e299643c7cb73a5a0f8b4d9beda621cb7e2f0ed481c349f3becb32ecb167b7
+size 4962221464
diff --git a/checkpoint-3000/model-00004-of-00004.safetensors b/checkpoint-3000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..1526271
--- /dev/null
+++ b/checkpoint-3000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e220d63db00d7a6c2a436202c5c0dfcc78e7e78dd99ffd2d2eba712f64a9add0
+size 3670322200
diff --git a/checkpoint-3000/model.safetensors.index.json b/checkpoint-3000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-3000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-3000/optimizer.pt b/checkpoint-3000/optimizer.pt
new file mode 100644
index 0000000..93e9945
--- /dev/null
+++ b/checkpoint-3000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c50396fa5781817d12afd00b901b848837d1b5799d4a807d0add8cbb8a16d221
+size 7927918951
diff --git a/checkpoint-3000/rng_state.pth b/checkpoint-3000/rng_state.pth
new file mode 100644
index 0000000..883145f
--- /dev/null
+++ b/checkpoint-3000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6bffe42dadda4047126a3977b014c9805b80cfac3ab223b376929dc323415be
+size 14645
diff --git a/checkpoint-3000/scheduler.pt b/checkpoint-3000/scheduler.pt
new file mode 100644
index 0000000..0394bb3
--- /dev/null
+++ b/checkpoint-3000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adf6915a7a446989bbb2f5defd721aa6c01a280fe6b24a77f753b0152041e35c
+size 1465
diff --git a/checkpoint-3000/special_tokens_map.json b/checkpoint-3000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-3000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-3000/tokenizer.json b/checkpoint-3000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-3000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-3000/tokenizer.model b/checkpoint-3000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-3000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-3000/tokenizer_config.json b/checkpoint-3000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-3000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-3000/trainer_state.json b/checkpoint-3000/trainer_state.json
new file mode 100644
index 0000000..a659a27
--- /dev/null
+++ b/checkpoint-3000/trainer_state.json
@@ -0,0 +1,3694 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.8,
+  "eval_steps": 100,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.514702006875267e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-3000/training_args.bin b/checkpoint-3000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-3000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-3500/chat_template.jinja b/checkpoint-3500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-3500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-3500/config.json b/checkpoint-3500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-3500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-3500/generation_config.json b/checkpoint-3500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-3500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-3500/model-00001-of-00004.safetensors b/checkpoint-3500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-3500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-3500/model-00002-of-00004.safetensors b/checkpoint-3500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-3500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-3500/model-00003-of-00004.safetensors b/checkpoint-3500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..5383528
--- /dev/null
+++ b/checkpoint-3500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4f32a12a4807d00d69b11dfc35b3d75842cebae8a7d2ae9e520e03a887975c7
+size 4962221464
diff --git a/checkpoint-3500/model-00004-of-00004.safetensors b/checkpoint-3500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..728b8ef
--- /dev/null
+++ b/checkpoint-3500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0f897a62efef139514bc5448e2c1f6f53ccafbc9621320a6999b45070599567e
+size 3670322200
diff --git a/checkpoint-3500/model.safetensors.index.json b/checkpoint-3500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-3500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-3500/optimizer.pt b/checkpoint-3500/optimizer.pt
new file mode 100644
index 0000000..eb0b0bb
--- /dev/null
+++ b/checkpoint-3500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72329c128586fcb8779d9bfee52e0d55b9d81c4714dd8b2b556b44da613e7336
+size 7927918951
diff --git a/checkpoint-3500/rng_state.pth b/checkpoint-3500/rng_state.pth
new file mode 100644
index 0000000..e452a45
--- /dev/null
+++ b/checkpoint-3500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b2ee99384404b1c9fced95b6f30834cff07f8fa00b35f7da820622ed257adfa
+size 14645
diff --git a/checkpoint-3500/scheduler.pt b/checkpoint-3500/scheduler.pt
new file mode 100644
index 0000000..3f1c6a3
--- /dev/null
+++ b/checkpoint-3500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff8615852849191877d2d8e131c28913ea0d29e591098da0946259116799c96e
+size 1465
diff --git a/checkpoint-3500/special_tokens_map.json b/checkpoint-3500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-3500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-3500/tokenizer.json b/checkpoint-3500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-3500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-3500/tokenizer.model b/checkpoint-3500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-3500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-3500/tokenizer_config.json b/checkpoint-3500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-3500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-3500/trainer_state.json b/checkpoint-3500/trainer_state.json
new file mode 100644
index 0000000..80cafd6
--- /dev/null
+++ b/checkpoint-3500/trainer_state.json
@@ -0,0 +1,4304 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.6,
+  "eval_steps": 100,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.600274972487193e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-3500/training_args.bin b/checkpoint-3500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-3500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-4000/chat_template.jinja b/checkpoint-4000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-4000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-4000/config.json b/checkpoint-4000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-4000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-4000/generation_config.json b/checkpoint-4000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-4000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-4000/model-00001-of-00004.safetensors b/checkpoint-4000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-4000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-4000/model-00002-of-00004.safetensors b/checkpoint-4000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-4000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-4000/model-00003-of-00004.safetensors b/checkpoint-4000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..ca86161
--- /dev/null
+++ b/checkpoint-4000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1ceae406c9264d8e186a5a9c948330798ac0b70783f4fa0f9990dac96a6d8711
+size 4962221464
diff --git a/checkpoint-4000/model-00004-of-00004.safetensors b/checkpoint-4000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..c82b5fb
--- /dev/null
+++ b/checkpoint-4000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f87ba68d28914e605563235dcfe7172ac8e95a17e0e5d49e1030c4857c93945
+size 3670322200
diff --git a/checkpoint-4000/model.safetensors.index.json b/checkpoint-4000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-4000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-4000/optimizer.pt b/checkpoint-4000/optimizer.pt
new file mode 100644
index 0000000..22b473b
--- /dev/null
+++ b/checkpoint-4000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba6fbf5cef2aa1df5048e364159c8104391917f2506619a5a09430578547fe00
+size 7927918951
diff --git a/checkpoint-4000/rng_state.pth b/checkpoint-4000/rng_state.pth
new file mode 100644
index 0000000..0400c9b
--- /dev/null
+++ b/checkpoint-4000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f0163768dee420fb57989ace79815cdde4fa8a8621cc9a84ac1b3d358ac5dfb
+size 14645
diff --git a/checkpoint-4000/scheduler.pt b/checkpoint-4000/scheduler.pt
new file mode 100644
index 0000000..e9387be
--- /dev/null
+++ b/checkpoint-4000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9927447b83167abe18564d4c4f3048aaac6ea3a59fdc776281f698b33bfa06a5
+size 1465
diff --git a/checkpoint-4000/special_tokens_map.json b/checkpoint-4000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-4000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-4000/tokenizer.json b/checkpoint-4000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-4000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-4000/tokenizer.model b/checkpoint-4000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-4000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-4000/tokenizer_config.json b/checkpoint-4000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-4000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-4000/trainer_state.json b/checkpoint-4000/trainer_state.json
new file mode 100644
index 0000000..76d62d9
--- /dev/null
+++ b/checkpoint-4000/trainer_state.json
@@ -0,0 +1,4914 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.4,
+  "eval_steps": 100,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.684832052488991e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-4000/training_args.bin b/checkpoint-4000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-4000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-4500/chat_template.jinja b/checkpoint-4500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-4500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-4500/config.json b/checkpoint-4500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-4500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-4500/generation_config.json b/checkpoint-4500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-4500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-4500/model-00001-of-00004.safetensors b/checkpoint-4500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-4500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-4500/model-00002-of-00004.safetensors b/checkpoint-4500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-4500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-4500/model-00003-of-00004.safetensors b/checkpoint-4500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..f383bbb
--- /dev/null
+++ b/checkpoint-4500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ab85f241e52c4c7b641d93320dbdb090807bba693a3c4a58ad5731782270e2f
+size 4962221464
diff --git a/checkpoint-4500/model-00004-of-00004.safetensors b/checkpoint-4500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..cb3313f
--- /dev/null
+++ b/checkpoint-4500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bb0e2a767d9d3ff74f33bdd606453b1efddd18d783536313c618a38dfe74bcf
+size 3670322200
diff --git a/checkpoint-4500/model.safetensors.index.json b/checkpoint-4500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-4500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-4500/optimizer.pt b/checkpoint-4500/optimizer.pt
new file mode 100644
index 0000000..b272b3b
--- /dev/null
+++ b/checkpoint-4500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f1f678637c9b1211b6e21191d96de283d9400d40a4d9958e9c4a0c3d8ffd7b83
+size 7927918951
diff --git a/checkpoint-4500/rng_state.pth b/checkpoint-4500/rng_state.pth
new file mode 100644
index 0000000..74ea931
--- /dev/null
+++ b/checkpoint-4500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:391a7dd508bb6cf0854fb5eef837bc3ae3b8c93ee720de83c3b9f9f6c298ea0a
+size 14645
diff --git a/checkpoint-4500/scheduler.pt b/checkpoint-4500/scheduler.pt
new file mode 100644
index 0000000..b3e692b
--- /dev/null
+++ b/checkpoint-4500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00e843f0ce9380e9aeae953b05f2acd9ab03c0d8559bdb44a4312f28a8513507
+size 1465
diff --git a/checkpoint-4500/special_tokens_map.json b/checkpoint-4500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-4500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-4500/tokenizer.json b/checkpoint-4500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-4500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-4500/tokenizer.model b/checkpoint-4500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-4500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-4500/tokenizer_config.json b/checkpoint-4500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-4500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-4500/trainer_state.json b/checkpoint-4500/trainer_state.json
new file mode 100644
index 0000000..68896ea
--- /dev/null
+++ b/checkpoint-4500/trainer_state.json
@@ -0,0 +1,5524 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.2,
+  "eval_steps": 100,
+  "global_step": 4500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 9.768063986128552e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-4500/training_args.bin b/checkpoint-4500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-4500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-500/chat_template.jinja b/checkpoint-500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-500/config.json b/checkpoint-500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-500/generation_config.json b/checkpoint-500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-500/model-00001-of-00004.safetensors b/checkpoint-500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-500/model-00002-of-00004.safetensors b/checkpoint-500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-500/model-00003-of-00004.safetensors b/checkpoint-500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..6df218f
--- /dev/null
+++ b/checkpoint-500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:56a2b3de2fb815156e39befade232902892224592e366bac54e7c67c14172dad
+size 4962221464
diff --git a/checkpoint-500/model-00004-of-00004.safetensors b/checkpoint-500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..d493f2e
--- /dev/null
+++ b/checkpoint-500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fb5fc3d9202aedc4a09bb2bbd63b3248fcafc11d07fc4215099f9b8fa46ef502
+size 3670322200
diff --git a/checkpoint-500/model.safetensors.index.json b/checkpoint-500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-500/optimizer.pt b/checkpoint-500/optimizer.pt
new file mode 100644
index 0000000..0749af9
--- /dev/null
+++ b/checkpoint-500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:94f019da9d99e1adb057f5cc63700cb81e2519a5c1f8382ee2479bb837803d56
+size 7927918951
diff --git a/checkpoint-500/rng_state.pth b/checkpoint-500/rng_state.pth
new file mode 100644
index 0000000..73a1a8c
--- /dev/null
+++ b/checkpoint-500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69b6509f8bbdd943c9f8984cdc19d5d17ca63b6e47cd6e28080ebf502da06ddc
+size 14645
diff --git a/checkpoint-500/scheduler.pt b/checkpoint-500/scheduler.pt
new file mode 100644
index 0000000..a186a2c
--- /dev/null
+++ b/checkpoint-500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63481e620bd287eefb582d20daf306c54c7268088fea6f19af8100945c7e3e7e
+size 1465
diff --git a/checkpoint-500/special_tokens_map.json b/checkpoint-500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-500/tokenizer.json b/checkpoint-500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-500/tokenizer.model b/checkpoint-500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-500/tokenizer_config.json b/checkpoint-500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-500/trainer_state.json b/checkpoint-500/trainer_state.json
new file mode 100644
index 0000000..c06c2da
--- /dev/null
+++ b/checkpoint-500/trainer_state.json
@@ -0,0 +1,644 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8,
+  "eval_steps": 100,
+  "global_step": 500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0863379264774472e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-500/training_args.bin b/checkpoint-500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-5000/chat_template.jinja b/checkpoint-5000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-5000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-5000/config.json b/checkpoint-5000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-5000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-5000/generation_config.json b/checkpoint-5000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-5000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-5000/model-00001-of-00004.safetensors b/checkpoint-5000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-5000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-5000/model-00002-of-00004.safetensors b/checkpoint-5000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-5000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-5000/model-00003-of-00004.safetensors b/checkpoint-5000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..0a7b03e
--- /dev/null
+++ b/checkpoint-5000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a48c63b560d1b3fffd66589a26f8686c93e6b9b8b7f7aa4d6b9cbc12dd38182
+size 4962221464
diff --git a/checkpoint-5000/model-00004-of-00004.safetensors b/checkpoint-5000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..9d1e205
--- /dev/null
+++ b/checkpoint-5000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28559d14f8e72f959127a19dd4546f81cc7c9b0ecbadddffb68f91ce1cbaac95
+size 3670322200
diff --git a/checkpoint-5000/model.safetensors.index.json b/checkpoint-5000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-5000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-5000/optimizer.pt b/checkpoint-5000/optimizer.pt
new file mode 100644
index 0000000..a13d446
--- /dev/null
+++ b/checkpoint-5000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0260d8b15e7efc84e8bdcf1c7ec80689cace2993deb17dbbe793973f286e03b
+size 7927918951
diff --git a/checkpoint-5000/rng_state.pth b/checkpoint-5000/rng_state.pth
new file mode 100644
index 0000000..06ec22d
--- /dev/null
+++ b/checkpoint-5000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c481ed78d3b96ffb2e3a019faa6b25af7640ce108d0549fb3d24fce7b2b2a0a
+size 14645
diff --git a/checkpoint-5000/scheduler.pt b/checkpoint-5000/scheduler.pt
new file mode 100644
index 0000000..f092dad
--- /dev/null
+++ b/checkpoint-5000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2d6fca2bf44137be134b72aa06167f7ba1f1b3d530be38a4b21e87094b1b63c
+size 1465
diff --git a/checkpoint-5000/special_tokens_map.json b/checkpoint-5000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-5000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-5000/tokenizer.json b/checkpoint-5000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-5000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-5000/tokenizer.model b/checkpoint-5000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-5000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-5000/tokenizer_config.json b/checkpoint-5000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-5000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-5000/trainer_state.json b/checkpoint-5000/trainer_state.json
new file mode 100644
index 0000000..1808775
--- /dev/null
+++ b/checkpoint-5000/trainer_state.json
@@ -0,0 +1,6134 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.0,
+  "eval_steps": 100,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.0853280143146635e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-5000/training_args.bin b/checkpoint-5000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-5000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-5500/chat_template.jinja b/checkpoint-5500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-5500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-5500/config.json b/checkpoint-5500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-5500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-5500/generation_config.json b/checkpoint-5500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-5500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-5500/model-00001-of-00004.safetensors b/checkpoint-5500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-5500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-5500/model-00002-of-00004.safetensors b/checkpoint-5500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-5500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-5500/model-00003-of-00004.safetensors b/checkpoint-5500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..fbffe5b
--- /dev/null
+++ b/checkpoint-5500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a4c37acd6ae7b44cfd17688912b873c0708aa498c60daadeed552f671622b2b1
+size 4962221464
diff --git a/checkpoint-5500/model-00004-of-00004.safetensors b/checkpoint-5500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..a8e0bf9
--- /dev/null
+++ b/checkpoint-5500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:af3974f728dd484ad8239cf79d0149994ac062b7e458daf7dc357b8f52c53647
+size 3670322200
diff --git a/checkpoint-5500/model.safetensors.index.json b/checkpoint-5500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-5500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-5500/optimizer.pt b/checkpoint-5500/optimizer.pt
new file mode 100644
index 0000000..da16afd
--- /dev/null
+++ b/checkpoint-5500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1fcdf2aa08c888015b05cb134cf60dead6b8a6e41ae03f5cfe9d428cad0ed36
+size 7927918951
diff --git a/checkpoint-5500/rng_state.pth b/checkpoint-5500/rng_state.pth
new file mode 100644
index 0000000..74c3844
--- /dev/null
+++ b/checkpoint-5500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8127923c344189b8a98802008e6aaf05c5f57af98bc4256519303177929b3ec3
+size 14645
diff --git a/checkpoint-5500/scheduler.pt b/checkpoint-5500/scheduler.pt
new file mode 100644
index 0000000..2f0aae7
--- /dev/null
+++ b/checkpoint-5500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff810b5bdc950a75b24c4ea305f384abe5f058cb9de8410ebddf3d7677c25b3a
+size 1465
diff --git a/checkpoint-5500/special_tokens_map.json b/checkpoint-5500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-5500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-5500/tokenizer.json b/checkpoint-5500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-5500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-5500/tokenizer.model b/checkpoint-5500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-5500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-5500/tokenizer_config.json b/checkpoint-5500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-5500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-5500/trainer_state.json b/checkpoint-5500/trainer_state.json
new file mode 100644
index 0000000..7b3f4a9
--- /dev/null
+++ b/checkpoint-5500/trainer_state.json
@@ -0,0 +1,6744 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 8.8,
+  "eval_steps": 100,
+  "global_step": 5500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.1938013629636878e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-5500/training_args.bin b/checkpoint-5500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-5500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-6000/chat_template.jinja b/checkpoint-6000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-6000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-6000/config.json b/checkpoint-6000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-6000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-6000/generation_config.json b/checkpoint-6000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-6000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-6000/model-00001-of-00004.safetensors b/checkpoint-6000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-6000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-6000/model-00002-of-00004.safetensors b/checkpoint-6000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-6000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-6000/model-00003-of-00004.safetensors b/checkpoint-6000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..ba30ee2
--- /dev/null
+++ b/checkpoint-6000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7af21334fa729c13e06bd0883822030fa1c5297e22e1b30737e0a61b6eb21c7
+size 4962221464
diff --git a/checkpoint-6000/model-00004-of-00004.safetensors b/checkpoint-6000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..6992d8b
--- /dev/null
+++ b/checkpoint-6000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77df8c4ffed82b705c76fda33c6db045731a61e333868f644122e60debc6dc39
+size 3670322200
diff --git a/checkpoint-6000/model.safetensors.index.json b/checkpoint-6000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-6000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-6000/optimizer.pt b/checkpoint-6000/optimizer.pt
new file mode 100644
index 0000000..aafd482
--- /dev/null
+++ b/checkpoint-6000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d17394e5c0bc30e22d1d76a39d862ee513ceeabe704afb2a6d2f6682776643be
+size 7927918951
diff --git a/checkpoint-6000/rng_state.pth b/checkpoint-6000/rng_state.pth
new file mode 100644
index 0000000..3486b9c
--- /dev/null
+++ b/checkpoint-6000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93efbcc63722fe4c84732b4bcf5673b509692665880e4b06672c2784005d758b
+size 14645
diff --git a/checkpoint-6000/scheduler.pt b/checkpoint-6000/scheduler.pt
new file mode 100644
index 0000000..2f927ab
--- /dev/null
+++ b/checkpoint-6000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e63ad7405b117dbeeed251777a7456b609dcb42d6fa093f584ab3971c2a5cfbf
+size 1465
diff --git a/checkpoint-6000/special_tokens_map.json b/checkpoint-6000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-6000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-6000/tokenizer.json b/checkpoint-6000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-6000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-6000/tokenizer.model b/checkpoint-6000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-6000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-6000/tokenizer_config.json b/checkpoint-6000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-6000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-6000/trainer_state.json b/checkpoint-6000/trainer_state.json
new file mode 100644
index 0000000..63f358a
--- /dev/null
+++ b/checkpoint-6000/trainer_state.json
@@ -0,0 +1,7354 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.6,
+  "eval_steps": 100,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.3026424681970037e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-6000/training_args.bin b/checkpoint-6000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-6000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-6500/chat_template.jinja b/checkpoint-6500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-6500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-6500/config.json b/checkpoint-6500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-6500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-6500/generation_config.json b/checkpoint-6500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-6500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-6500/model-00001-of-00004.safetensors b/checkpoint-6500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-6500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-6500/model-00002-of-00004.safetensors b/checkpoint-6500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-6500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-6500/model-00003-of-00004.safetensors b/checkpoint-6500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..5e4e573
--- /dev/null
+++ b/checkpoint-6500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:df779ef4df77980cdd2f6d2cae4079354604bc38b6ace87c1a9a41e7e6db85d7
+size 4962221464
diff --git a/checkpoint-6500/model-00004-of-00004.safetensors b/checkpoint-6500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..e09b194
--- /dev/null
+++ b/checkpoint-6500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e8e9aa4118d22c678d12613ab9e5fe7d29831a59272fc4cd491746e2156b3b1
+size 3670322200
diff --git a/checkpoint-6500/model.safetensors.index.json b/checkpoint-6500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-6500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-6500/optimizer.pt b/checkpoint-6500/optimizer.pt
new file mode 100644
index 0000000..4309402
--- /dev/null
+++ b/checkpoint-6500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09cfd0f8e92efa5c8adb06274fceea627cacc36c90d07594ca5f84bc5c84736f
+size 7927918951
diff --git a/checkpoint-6500/rng_state.pth b/checkpoint-6500/rng_state.pth
new file mode 100644
index 0000000..9dfea11
--- /dev/null
+++ b/checkpoint-6500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9230cf3c20a145ef73f101953b2f9f1d6f54b18d7c135f24b78eec56a412431c
+size 14645
diff --git a/checkpoint-6500/scheduler.pt b/checkpoint-6500/scheduler.pt
new file mode 100644
index 0000000..35008fa
--- /dev/null
+++ b/checkpoint-6500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb78d975548f28d0aee6fd506c9f5c5346c2d1f177048b058e5c909ed97941e9
+size 1465
diff --git a/checkpoint-6500/special_tokens_map.json b/checkpoint-6500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-6500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-6500/tokenizer.json b/checkpoint-6500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-6500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-6500/tokenizer.model b/checkpoint-6500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-6500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-6500/tokenizer_config.json b/checkpoint-6500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-6500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-6500/trainer_state.json b/checkpoint-6500/trainer_state.json
new file mode 100644
index 0000000..6f56e4a
--- /dev/null
+++ b/checkpoint-6500/trainer_state.json
@@ -0,0 +1,7964 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.4,
+  "eval_steps": 100,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4106312284220506e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-6500/training_args.bin b/checkpoint-6500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-6500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-7000/chat_template.jinja b/checkpoint-7000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-7000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-7000/config.json b/checkpoint-7000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-7000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-7000/generation_config.json b/checkpoint-7000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-7000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-7000/model-00001-of-00004.safetensors b/checkpoint-7000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-7000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-7000/model-00002-of-00004.safetensors b/checkpoint-7000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-7000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-7000/model-00003-of-00004.safetensors b/checkpoint-7000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..7db02f2
--- /dev/null
+++ b/checkpoint-7000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a141663c26a2aa3337fa021b40305f2453b91c8bd9cbe671441db1d1a6c6a32f
+size 4962221464
diff --git a/checkpoint-7000/model-00004-of-00004.safetensors b/checkpoint-7000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..586c1f9
--- /dev/null
+++ b/checkpoint-7000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca918778c54d7ce28fac53d78c1611e448c3974267c08b5823f5e6e46385d259
+size 3670322200
diff --git a/checkpoint-7000/model.safetensors.index.json b/checkpoint-7000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-7000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-7000/optimizer.pt b/checkpoint-7000/optimizer.pt
new file mode 100644
index 0000000..033f5b8
--- /dev/null
+++ b/checkpoint-7000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:20ab0b13f46c29afab0c95856e9de839b8e82c9c82d1bdb49170eb5121aa6e22
+size 7927918951
diff --git a/checkpoint-7000/rng_state.pth b/checkpoint-7000/rng_state.pth
new file mode 100644
index 0000000..bb16bf3
--- /dev/null
+++ b/checkpoint-7000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:99f794db99fe07805f4569ccba2eadce029179c6a0e2e7b91b66575531731edf
+size 14645
diff --git a/checkpoint-7000/scheduler.pt b/checkpoint-7000/scheduler.pt
new file mode 100644
index 0000000..e05f089
--- /dev/null
+++ b/checkpoint-7000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18c76915fc982ac3ab524a84f455e6d21c6a8b6dace12db05f52eb2b022aa4b3
+size 1465
diff --git a/checkpoint-7000/special_tokens_map.json b/checkpoint-7000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-7000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-7000/tokenizer.json b/checkpoint-7000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-7000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-7000/tokenizer.model b/checkpoint-7000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-7000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-7000/tokenizer_config.json b/checkpoint-7000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-7000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-7000/trainer_state.json b/checkpoint-7000/trainer_state.json
new file mode 100644
index 0000000..69f978d
--- /dev/null
+++ b/checkpoint-7000/trainer_state.json
@@ -0,0 +1,8574 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 11.2,
+  "eval_steps": 100,
+  "global_step": 7000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.5188036072241414e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-7000/training_args.bin b/checkpoint-7000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-7000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-7500/chat_template.jinja b/checkpoint-7500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-7500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-7500/config.json b/checkpoint-7500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-7500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-7500/generation_config.json b/checkpoint-7500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-7500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-7500/model-00001-of-00004.safetensors b/checkpoint-7500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-7500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-7500/model-00002-of-00004.safetensors b/checkpoint-7500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-7500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-7500/model-00003-of-00004.safetensors b/checkpoint-7500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..dc6d385
--- /dev/null
+++ b/checkpoint-7500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3249c680288ea576aa2a5f34ce1894f8d966c36ddaf44c3713c28b011a51a429
+size 4962221464
diff --git a/checkpoint-7500/model-00004-of-00004.safetensors b/checkpoint-7500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..e50fee4
--- /dev/null
+++ b/checkpoint-7500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:63ad7e29deb252ffc268c8a992927abbc901dcb383de8e68874600134d95db5b
+size 3670322200
diff --git a/checkpoint-7500/model.safetensors.index.json b/checkpoint-7500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-7500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-7500/optimizer.pt b/checkpoint-7500/optimizer.pt
new file mode 100644
index 0000000..40ab17e
--- /dev/null
+++ b/checkpoint-7500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7c3976d19427072f0d6aa56b532e4c674048e9a95efd795d72324af99c8f113b
+size 7927918951
diff --git a/checkpoint-7500/rng_state.pth b/checkpoint-7500/rng_state.pth
new file mode 100644
index 0000000..8e9afa1
--- /dev/null
+++ b/checkpoint-7500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d2bd7ee1febe876510cfeafa6c69645c4c036aea4ea713eef86eee9127945e99
+size 14645
diff --git a/checkpoint-7500/scheduler.pt b/checkpoint-7500/scheduler.pt
new file mode 100644
index 0000000..ed219df
--- /dev/null
+++ b/checkpoint-7500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f64aac8779ecc92383de34b62a43130dde40d000c64dfb79667fd7cda6581c6
+size 1465
diff --git a/checkpoint-7500/special_tokens_map.json b/checkpoint-7500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-7500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-7500/tokenizer.json b/checkpoint-7500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-7500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-7500/tokenizer.model b/checkpoint-7500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-7500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-7500/tokenizer_config.json b/checkpoint-7500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-7500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-7500/trainer_state.json b/checkpoint-7500/trainer_state.json
new file mode 100644
index 0000000..8024091
--- /dev/null
+++ b/checkpoint-7500/trainer_state.json
@@ -0,0 +1,9184 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.0,
+  "eval_steps": 100,
+  "global_step": 7500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.6276305879515308e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-7500/training_args.bin b/checkpoint-7500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-7500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-8000/chat_template.jinja b/checkpoint-8000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-8000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-8000/config.json b/checkpoint-8000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-8000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-8000/generation_config.json b/checkpoint-8000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-8000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-8000/model-00001-of-00004.safetensors b/checkpoint-8000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-8000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-8000/model-00002-of-00004.safetensors b/checkpoint-8000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-8000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-8000/model-00003-of-00004.safetensors b/checkpoint-8000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..b137e4a
--- /dev/null
+++ b/checkpoint-8000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2423ce5c29f871b1af82cf7ad5558c4de6a694ac180879b2c3ad81650832be32
+size 4962221464
diff --git a/checkpoint-8000/model-00004-of-00004.safetensors b/checkpoint-8000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..5dfddbf
--- /dev/null
+++ b/checkpoint-8000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cec7774154144f0f435c9c6cdf0878614221a4848558ba7bcd426c1d52300ec3
+size 3670322200
diff --git a/checkpoint-8000/model.safetensors.index.json b/checkpoint-8000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-8000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-8000/optimizer.pt b/checkpoint-8000/optimizer.pt
new file mode 100644
index 0000000..68ebd61
--- /dev/null
+++ b/checkpoint-8000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2e54ddee2b9e64fa9242f60091cd685c17843e76bf9a5c1e22b007668aa71e8d
+size 7927918951
diff --git a/checkpoint-8000/rng_state.pth b/checkpoint-8000/rng_state.pth
new file mode 100644
index 0000000..6c99c3f
--- /dev/null
+++ b/checkpoint-8000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeed9af5dc6675410383fbdeb58b778d8a04c8989ca53bc40c26d41b58280bfa
+size 14645
diff --git a/checkpoint-8000/scheduler.pt b/checkpoint-8000/scheduler.pt
new file mode 100644
index 0000000..9acaa85
--- /dev/null
+++ b/checkpoint-8000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c20d4ef0f6cfa0b5bfbbfc40fb5bd0e3c14a6b2cc6eda7232275e58d1ea2452f
+size 1465
diff --git a/checkpoint-8000/special_tokens_map.json b/checkpoint-8000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-8000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-8000/tokenizer.json b/checkpoint-8000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-8000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-8000/tokenizer.model b/checkpoint-8000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-8000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-8000/tokenizer_config.json b/checkpoint-8000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-8000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-8000/trainer_state.json b/checkpoint-8000/trainer_state.json
new file mode 100644
index 0000000..ee79218
--- /dev/null
+++ b/checkpoint-8000/trainer_state.json
@@ -0,0 +1,9794 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 12.8,
+  "eval_steps": 100,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    },
+    {
+      "entropy": 0.053174392925575374,
+      "epoch": 12.016,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.535555555555556e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905574269592762,
+      "num_tokens": 214920834.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.05121934600174427,
+      "epoch": 12.032,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.513333333333334e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907024279236794,
+      "num_tokens": 215207212.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.05282459647860378,
+      "epoch": 12.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.491111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904284842312336,
+      "num_tokens": 215491551.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.0523437969153747,
+      "epoch": 12.064,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.468888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907445590943098,
+      "num_tokens": 215773788.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.05109455803176388,
+      "epoch": 12.08,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990898734331131,
+      "num_tokens": 216063342.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.052485697087831795,
+      "epoch": 12.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.424444444444444e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905819371342659,
+      "num_tokens": 216346341.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.05154973501339555,
+      "epoch": 12.112,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.402222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906920839101077,
+      "num_tokens": 216631586.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.05027051754295826,
+      "epoch": 12.128,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908485073596239,
+      "num_tokens": 216921387.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.051271379156969485,
+      "epoch": 12.144,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.357777777777779e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907938674092293,
+      "num_tokens": 217208948.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.05280535111669451,
+      "epoch": 12.16,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904177527874708,
+      "num_tokens": 217488087.0,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_math_entropy": 0.17994125121831894,
+      "eval_math_loss": 1.3322224617004395,
+      "eval_math_mean_token_accuracy": 0.8160332980155944,
+      "eval_math_num_tokens": 217488087.0,
+      "eval_math_runtime": 35.8103,
+      "eval_math_samples_per_second": 13.962,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_physics_entropy": 0.20668217623233795,
+      "eval_physics_loss": 1.3987233638763428,
+      "eval_physics_mean_token_accuracy": 0.7963971982002258,
+      "eval_physics_num_tokens": 217488087.0,
+      "eval_physics_runtime": 41.4958,
+      "eval_physics_samples_per_second": 12.049,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7600
+    },
+    {
+      "entropy": 0.05196192401926965,
+      "epoch": 12.176,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.313333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906202860176563,
+      "num_tokens": 217770038.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.051902253658045086,
+      "epoch": 12.192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.2911111111111115e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9906028714030981,
+      "num_tokens": 218050486.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.052790193830151114,
+      "epoch": 12.208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.268888888888889e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904781151562929,
+      "num_tokens": 218331465.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.052298348920885473,
+      "epoch": 12.224,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.246666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906631808727979,
+      "num_tokens": 218618801.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.050798607314936815,
+      "epoch": 12.24,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9909202449023724,
+      "num_tokens": 218910596.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.052114113536663356,
+      "epoch": 12.256,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.202222222222223e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905635304749012,
+      "num_tokens": 219194842.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.05205410345224663,
+      "epoch": 12.272,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990602058544755,
+      "num_tokens": 219484076.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.049987882829736915,
+      "epoch": 12.288,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.157777777777778e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909869834780693,
+      "num_tokens": 219780222.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.05258604409173131,
+      "epoch": 12.304,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.135555555555556e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904522281140089,
+      "num_tokens": 220065272.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.05357218148419633,
+      "epoch": 12.32,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9902280792593956,
+      "num_tokens": 220344110.0,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_math_entropy": 0.1793700819015503,
+      "eval_math_loss": 1.3329575061798096,
+      "eval_math_mean_token_accuracy": 0.8161038937568664,
+      "eval_math_num_tokens": 220344110.0,
+      "eval_math_runtime": 35.8174,
+      "eval_math_samples_per_second": 13.96,
+      "eval_math_steps_per_second": 3.49,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_physics_entropy": 0.20628326386213303,
+      "eval_physics_loss": 1.3988101482391357,
+      "eval_physics_mean_token_accuracy": 0.7963225340843201,
+      "eval_physics_num_tokens": 220344110.0,
+      "eval_physics_runtime": 41.5047,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7700
+    },
+    {
+      "entropy": 0.05083432825049385,
+      "epoch": 12.336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.091111111111111e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906401328742505,
+      "num_tokens": 220632516.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.05188987164292484,
+      "epoch": 12.352,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.06888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905499298125505,
+      "num_tokens": 220918123.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.05209730467759073,
+      "epoch": 12.368,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.046666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905961699783802,
+      "num_tokens": 221206063.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.05238552400842309,
+      "epoch": 12.384,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.024444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904579646885395,
+      "num_tokens": 221488317.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.05181863873731345,
+      "epoch": 12.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905485171824694,
+      "num_tokens": 221776008.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.05192954689264297,
+      "epoch": 12.416,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904752410948277,
+      "num_tokens": 222057789.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.052794202556833626,
+      "epoch": 12.432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.957777777777778e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990259325504303,
+      "num_tokens": 222339915.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.052391084167174994,
+      "epoch": 12.448,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903523031622171,
+      "num_tokens": 222619687.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.05029154985677451,
+      "epoch": 12.464,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.9133333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906887255609036,
+      "num_tokens": 222913264.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.05082799810916185,
+      "epoch": 12.48,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906626105308532,
+      "num_tokens": 223202947.0,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_math_entropy": 0.180211255133152,
+      "eval_math_loss": 1.333121657371521,
+      "eval_math_mean_token_accuracy": 0.8160158872604371,
+      "eval_math_num_tokens": 223202947.0,
+      "eval_math_runtime": 35.8241,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_physics_entropy": 0.20699969559907913,
+      "eval_physics_loss": 1.3986775875091553,
+      "eval_physics_mean_token_accuracy": 0.7963728904724121,
+      "eval_physics_num_tokens": 223202947.0,
+      "eval_physics_runtime": 41.4724,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 7800
+    },
+    {
+      "entropy": 0.05148934966418892,
+      "epoch": 12.496,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.8688888888888895e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905773665755987,
+      "num_tokens": 223491040.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.05243352035759017,
+      "epoch": 12.512,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.846666666666667e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903561118990183,
+      "num_tokens": 223775825.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.05058162130881101,
+      "epoch": 12.528,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.824444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990830871462822,
+      "num_tokens": 224068092.0,
+      "step": 7830
+    },
+    {
+      "entropy": 0.051722967089153825,
+      "epoch": 12.544,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.802222222222222e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903257399797439,
+      "num_tokens": 224354624.0,
+      "step": 7840
+    },
+    {
+      "entropy": 0.05123099538031965,
+      "epoch": 12.56,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906783111393451,
+      "num_tokens": 224641974.0,
+      "step": 7850
+    },
+    {
+      "entropy": 0.05136660183779895,
+      "epoch": 12.576,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.7577777777777784e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906629983335733,
+      "num_tokens": 224932014.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.05145444863010198,
+      "epoch": 12.592,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.735555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9903172593563795,
+      "num_tokens": 225222434.0,
+      "step": 7870
+    },
+    {
+      "entropy": 0.05075638788985089,
+      "epoch": 12.608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.713333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906791742891073,
+      "num_tokens": 225515803.0,
+      "step": 7880
+    },
+    {
+      "entropy": 0.05076117676217109,
+      "epoch": 12.624,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.691111111111111e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906085334718228,
+      "num_tokens": 225810131.0,
+      "step": 7890
+    },
+    {
+      "entropy": 0.05141493146074936,
+      "epoch": 12.64,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904979523271322,
+      "num_tokens": 226100172.0,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_math_entropy": 0.17927622705698013,
+      "eval_math_loss": 1.332814335823059,
+      "eval_math_mean_token_accuracy": 0.8159863357543945,
+      "eval_math_num_tokens": 226100172.0,
+      "eval_math_runtime": 35.8058,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_physics_entropy": 0.20605486339330673,
+      "eval_physics_loss": 1.3984746932983398,
+      "eval_physics_mean_token_accuracy": 0.7964399604797363,
+      "eval_physics_num_tokens": 226100172.0,
+      "eval_physics_runtime": 41.5077,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7900
+    },
+    {
+      "entropy": 0.052684175770264116,
+      "epoch": 12.656,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.646666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902132380753755,
+      "num_tokens": 226379811.0,
+      "step": 7910
+    },
+    {
+      "entropy": 0.05348572533112019,
+      "epoch": 12.672,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.990258751064539,
+      "num_tokens": 226662085.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.05106104960432276,
+      "epoch": 12.688,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.602222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9902874868363142,
+      "num_tokens": 226950257.0,
+      "step": 7930
+    },
+    {
+      "entropy": 0.0518765551969409,
+      "epoch": 12.704,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905543696135283,
+      "num_tokens": 227238669.0,
+      "step": 7940
+    },
+    {
+      "entropy": 0.05253795627504587,
+      "epoch": 12.72,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990134946629405,
+      "num_tokens": 227518695.0,
+      "step": 7950
+    },
+    {
+      "entropy": 0.05207758563337848,
+      "epoch": 12.736,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.535555555555555e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990263070166111,
+      "num_tokens": 227806156.0,
+      "step": 7960
+    },
+    {
+      "entropy": 0.05179664236493409,
+      "epoch": 12.752,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.513333333333333e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904587935656309,
+      "num_tokens": 228091641.0,
+      "step": 7970
+    },
+    {
+      "entropy": 0.05190181992948055,
+      "epoch": 12.768,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.4911111111111115e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9901351545006036,
+      "num_tokens": 228376952.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.0528141722548753,
+      "epoch": 12.784,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.468888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901187218725681,
+      "num_tokens": 228657003.0,
+      "step": 7990
+    },
+    {
+      "entropy": 0.051391151722054926,
+      "epoch": 12.8,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.446666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9902074325829744,
+      "num_tokens": 228942799.0,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_math_entropy": 0.1799713587164879,
+      "eval_math_loss": 1.332999587059021,
+      "eval_math_mean_token_accuracy": 0.8159932594299316,
+      "eval_math_num_tokens": 228942799.0,
+      "eval_math_runtime": 35.8043,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_physics_entropy": 0.20658580523729325,
+      "eval_physics_loss": 1.3991401195526123,
+      "eval_physics_mean_token_accuracy": 0.7965000495910645,
+      "eval_physics_num_tokens": 228942799.0,
+      "eval_physics_runtime": 41.1662,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.736015373750807e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-8000/training_args.bin b/checkpoint-8000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-8000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-8500/chat_template.jinja b/checkpoint-8500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-8500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-8500/config.json b/checkpoint-8500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-8500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-8500/generation_config.json b/checkpoint-8500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-8500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-8500/model-00001-of-00004.safetensors b/checkpoint-8500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-8500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-8500/model-00002-of-00004.safetensors b/checkpoint-8500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-8500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-8500/model-00003-of-00004.safetensors b/checkpoint-8500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..6f4d87b
--- /dev/null
+++ b/checkpoint-8500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ae25f16eb7f960ef0808071564a421818be2a92c31b07210b4158fad4658d89
+size 4962221464
diff --git a/checkpoint-8500/model-00004-of-00004.safetensors b/checkpoint-8500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..007e222
--- /dev/null
+++ b/checkpoint-8500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fca268bee8eea54ad97d0c0cfdae9fcf5c55018527e8e84f8b96caee8fa0882
+size 3670322200
diff --git a/checkpoint-8500/model.safetensors.index.json b/checkpoint-8500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-8500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-8500/optimizer.pt b/checkpoint-8500/optimizer.pt
new file mode 100644
index 0000000..585a7e7
--- /dev/null
+++ b/checkpoint-8500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c19a7d189bff08bce6e4f990c6cc137d96e8f1f0297c1bc38f44dd257bb5d9d1
+size 7927918951
diff --git a/checkpoint-8500/rng_state.pth b/checkpoint-8500/rng_state.pth
new file mode 100644
index 0000000..86d7954
--- /dev/null
+++ b/checkpoint-8500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bcc286ff93068af3b7eefffeedce10c7fbcb9ce4ed1f27a87848ac8e96d92d3a
+size 14645
diff --git a/checkpoint-8500/scheduler.pt b/checkpoint-8500/scheduler.pt
new file mode 100644
index 0000000..b8eb29c
--- /dev/null
+++ b/checkpoint-8500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e79a64b90deea48fdbe90dd14a49a91c4ecbfd65b5e36abfcabc26f335a63781
+size 1465
diff --git a/checkpoint-8500/special_tokens_map.json b/checkpoint-8500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-8500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-8500/tokenizer.json b/checkpoint-8500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-8500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-8500/tokenizer.model b/checkpoint-8500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-8500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-8500/tokenizer_config.json b/checkpoint-8500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-8500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-8500/trainer_state.json b/checkpoint-8500/trainer_state.json
new file mode 100644
index 0000000..75ea81d
--- /dev/null
+++ b/checkpoint-8500/trainer_state.json
@@ -0,0 +1,10404 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 13.6,
+  "eval_steps": 100,
+  "global_step": 8500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    },
+    {
+      "entropy": 0.053174392925575374,
+      "epoch": 12.016,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.535555555555556e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905574269592762,
+      "num_tokens": 214920834.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.05121934600174427,
+      "epoch": 12.032,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.513333333333334e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907024279236794,
+      "num_tokens": 215207212.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.05282459647860378,
+      "epoch": 12.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.491111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904284842312336,
+      "num_tokens": 215491551.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.0523437969153747,
+      "epoch": 12.064,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.468888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907445590943098,
+      "num_tokens": 215773788.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.05109455803176388,
+      "epoch": 12.08,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990898734331131,
+      "num_tokens": 216063342.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.052485697087831795,
+      "epoch": 12.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.424444444444444e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905819371342659,
+      "num_tokens": 216346341.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.05154973501339555,
+      "epoch": 12.112,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.402222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906920839101077,
+      "num_tokens": 216631586.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.05027051754295826,
+      "epoch": 12.128,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908485073596239,
+      "num_tokens": 216921387.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.051271379156969485,
+      "epoch": 12.144,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.357777777777779e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907938674092293,
+      "num_tokens": 217208948.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.05280535111669451,
+      "epoch": 12.16,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904177527874708,
+      "num_tokens": 217488087.0,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_math_entropy": 0.17994125121831894,
+      "eval_math_loss": 1.3322224617004395,
+      "eval_math_mean_token_accuracy": 0.8160332980155944,
+      "eval_math_num_tokens": 217488087.0,
+      "eval_math_runtime": 35.8103,
+      "eval_math_samples_per_second": 13.962,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_physics_entropy": 0.20668217623233795,
+      "eval_physics_loss": 1.3987233638763428,
+      "eval_physics_mean_token_accuracy": 0.7963971982002258,
+      "eval_physics_num_tokens": 217488087.0,
+      "eval_physics_runtime": 41.4958,
+      "eval_physics_samples_per_second": 12.049,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7600
+    },
+    {
+      "entropy": 0.05196192401926965,
+      "epoch": 12.176,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.313333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906202860176563,
+      "num_tokens": 217770038.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.051902253658045086,
+      "epoch": 12.192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.2911111111111115e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9906028714030981,
+      "num_tokens": 218050486.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.052790193830151114,
+      "epoch": 12.208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.268888888888889e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904781151562929,
+      "num_tokens": 218331465.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.052298348920885473,
+      "epoch": 12.224,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.246666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906631808727979,
+      "num_tokens": 218618801.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.050798607314936815,
+      "epoch": 12.24,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9909202449023724,
+      "num_tokens": 218910596.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.052114113536663356,
+      "epoch": 12.256,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.202222222222223e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905635304749012,
+      "num_tokens": 219194842.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.05205410345224663,
+      "epoch": 12.272,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990602058544755,
+      "num_tokens": 219484076.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.049987882829736915,
+      "epoch": 12.288,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.157777777777778e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909869834780693,
+      "num_tokens": 219780222.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.05258604409173131,
+      "epoch": 12.304,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.135555555555556e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904522281140089,
+      "num_tokens": 220065272.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.05357218148419633,
+      "epoch": 12.32,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9902280792593956,
+      "num_tokens": 220344110.0,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_math_entropy": 0.1793700819015503,
+      "eval_math_loss": 1.3329575061798096,
+      "eval_math_mean_token_accuracy": 0.8161038937568664,
+      "eval_math_num_tokens": 220344110.0,
+      "eval_math_runtime": 35.8174,
+      "eval_math_samples_per_second": 13.96,
+      "eval_math_steps_per_second": 3.49,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_physics_entropy": 0.20628326386213303,
+      "eval_physics_loss": 1.3988101482391357,
+      "eval_physics_mean_token_accuracy": 0.7963225340843201,
+      "eval_physics_num_tokens": 220344110.0,
+      "eval_physics_runtime": 41.5047,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7700
+    },
+    {
+      "entropy": 0.05083432825049385,
+      "epoch": 12.336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.091111111111111e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906401328742505,
+      "num_tokens": 220632516.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.05188987164292484,
+      "epoch": 12.352,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.06888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905499298125505,
+      "num_tokens": 220918123.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.05209730467759073,
+      "epoch": 12.368,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.046666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905961699783802,
+      "num_tokens": 221206063.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.05238552400842309,
+      "epoch": 12.384,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.024444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904579646885395,
+      "num_tokens": 221488317.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.05181863873731345,
+      "epoch": 12.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905485171824694,
+      "num_tokens": 221776008.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.05192954689264297,
+      "epoch": 12.416,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904752410948277,
+      "num_tokens": 222057789.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.052794202556833626,
+      "epoch": 12.432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.957777777777778e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990259325504303,
+      "num_tokens": 222339915.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.052391084167174994,
+      "epoch": 12.448,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903523031622171,
+      "num_tokens": 222619687.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.05029154985677451,
+      "epoch": 12.464,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.9133333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906887255609036,
+      "num_tokens": 222913264.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.05082799810916185,
+      "epoch": 12.48,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906626105308532,
+      "num_tokens": 223202947.0,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_math_entropy": 0.180211255133152,
+      "eval_math_loss": 1.333121657371521,
+      "eval_math_mean_token_accuracy": 0.8160158872604371,
+      "eval_math_num_tokens": 223202947.0,
+      "eval_math_runtime": 35.8241,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_physics_entropy": 0.20699969559907913,
+      "eval_physics_loss": 1.3986775875091553,
+      "eval_physics_mean_token_accuracy": 0.7963728904724121,
+      "eval_physics_num_tokens": 223202947.0,
+      "eval_physics_runtime": 41.4724,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 7800
+    },
+    {
+      "entropy": 0.05148934966418892,
+      "epoch": 12.496,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.8688888888888895e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905773665755987,
+      "num_tokens": 223491040.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.05243352035759017,
+      "epoch": 12.512,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.846666666666667e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903561118990183,
+      "num_tokens": 223775825.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.05058162130881101,
+      "epoch": 12.528,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.824444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990830871462822,
+      "num_tokens": 224068092.0,
+      "step": 7830
+    },
+    {
+      "entropy": 0.051722967089153825,
+      "epoch": 12.544,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.802222222222222e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903257399797439,
+      "num_tokens": 224354624.0,
+      "step": 7840
+    },
+    {
+      "entropy": 0.05123099538031965,
+      "epoch": 12.56,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906783111393451,
+      "num_tokens": 224641974.0,
+      "step": 7850
+    },
+    {
+      "entropy": 0.05136660183779895,
+      "epoch": 12.576,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.7577777777777784e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906629983335733,
+      "num_tokens": 224932014.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.05145444863010198,
+      "epoch": 12.592,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.735555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9903172593563795,
+      "num_tokens": 225222434.0,
+      "step": 7870
+    },
+    {
+      "entropy": 0.05075638788985089,
+      "epoch": 12.608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.713333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906791742891073,
+      "num_tokens": 225515803.0,
+      "step": 7880
+    },
+    {
+      "entropy": 0.05076117676217109,
+      "epoch": 12.624,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.691111111111111e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906085334718228,
+      "num_tokens": 225810131.0,
+      "step": 7890
+    },
+    {
+      "entropy": 0.05141493146074936,
+      "epoch": 12.64,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904979523271322,
+      "num_tokens": 226100172.0,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_math_entropy": 0.17927622705698013,
+      "eval_math_loss": 1.332814335823059,
+      "eval_math_mean_token_accuracy": 0.8159863357543945,
+      "eval_math_num_tokens": 226100172.0,
+      "eval_math_runtime": 35.8058,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_physics_entropy": 0.20605486339330673,
+      "eval_physics_loss": 1.3984746932983398,
+      "eval_physics_mean_token_accuracy": 0.7964399604797363,
+      "eval_physics_num_tokens": 226100172.0,
+      "eval_physics_runtime": 41.5077,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7900
+    },
+    {
+      "entropy": 0.052684175770264116,
+      "epoch": 12.656,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.646666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902132380753755,
+      "num_tokens": 226379811.0,
+      "step": 7910
+    },
+    {
+      "entropy": 0.05348572533112019,
+      "epoch": 12.672,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.990258751064539,
+      "num_tokens": 226662085.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.05106104960432276,
+      "epoch": 12.688,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.602222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9902874868363142,
+      "num_tokens": 226950257.0,
+      "step": 7930
+    },
+    {
+      "entropy": 0.0518765551969409,
+      "epoch": 12.704,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905543696135283,
+      "num_tokens": 227238669.0,
+      "step": 7940
+    },
+    {
+      "entropy": 0.05253795627504587,
+      "epoch": 12.72,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990134946629405,
+      "num_tokens": 227518695.0,
+      "step": 7950
+    },
+    {
+      "entropy": 0.05207758563337848,
+      "epoch": 12.736,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.535555555555555e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990263070166111,
+      "num_tokens": 227806156.0,
+      "step": 7960
+    },
+    {
+      "entropy": 0.05179664236493409,
+      "epoch": 12.752,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.513333333333333e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904587935656309,
+      "num_tokens": 228091641.0,
+      "step": 7970
+    },
+    {
+      "entropy": 0.05190181992948055,
+      "epoch": 12.768,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.4911111111111115e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9901351545006036,
+      "num_tokens": 228376952.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.0528141722548753,
+      "epoch": 12.784,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.468888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901187218725681,
+      "num_tokens": 228657003.0,
+      "step": 7990
+    },
+    {
+      "entropy": 0.051391151722054926,
+      "epoch": 12.8,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.446666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9902074325829744,
+      "num_tokens": 228942799.0,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_math_entropy": 0.1799713587164879,
+      "eval_math_loss": 1.332999587059021,
+      "eval_math_mean_token_accuracy": 0.8159932594299316,
+      "eval_math_num_tokens": 228942799.0,
+      "eval_math_runtime": 35.8043,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_physics_entropy": 0.20658580523729325,
+      "eval_physics_loss": 1.3991401195526123,
+      "eval_physics_mean_token_accuracy": 0.7965000495910645,
+      "eval_physics_num_tokens": 228942799.0,
+      "eval_physics_runtime": 41.1662,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 8000
+    },
+    {
+      "entropy": 0.05222373401047662,
+      "epoch": 12.816,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.424444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902573928236962,
+      "num_tokens": 229230267.0,
+      "step": 8010
+    },
+    {
+      "entropy": 0.05221812457311899,
+      "epoch": 12.832,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.402222222222223e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902742177248001,
+      "num_tokens": 229513807.0,
+      "step": 8020
+    },
+    {
+      "entropy": 0.05192019739188254,
+      "epoch": 12.848,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990331218764186,
+      "num_tokens": 229800116.0,
+      "step": 8030
+    },
+    {
+      "entropy": 0.05204707682132721,
+      "epoch": 12.864,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904342792928219,
+      "num_tokens": 230086560.0,
+      "step": 8040
+    },
+    {
+      "entropy": 0.05261244847206399,
+      "epoch": 12.88,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.3355555555555565e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901342626661063,
+      "num_tokens": 230370705.0,
+      "step": 8050
+    },
+    {
+      "entropy": 0.050930758623871954,
+      "epoch": 12.896,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.313333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904414359480143,
+      "num_tokens": 230665090.0,
+      "step": 8060
+    },
+    {
+      "entropy": 0.0521988584427163,
+      "epoch": 12.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.291111111111112e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902524299919605,
+      "num_tokens": 230948774.0,
+      "step": 8070
+    },
+    {
+      "entropy": 0.05130416583269835,
+      "epoch": 12.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.268888888888889e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902212977409363,
+      "num_tokens": 231236122.0,
+      "step": 8080
+    },
+    {
+      "entropy": 0.05124916716013104,
+      "epoch": 12.943999999999999,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.246666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990312222391367,
+      "num_tokens": 231520994.0,
+      "step": 8090
+    },
+    {
+      "entropy": 0.05303966203937307,
+      "epoch": 12.96,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.2244444444444446e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9900147426873446,
+      "num_tokens": 231803509.0,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_math_entropy": 0.18036945784091948,
+      "eval_math_loss": 1.332650899887085,
+      "eval_math_mean_token_accuracy": 0.8161214051246644,
+      "eval_math_num_tokens": 231803509.0,
+      "eval_math_runtime": 35.8002,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_physics_entropy": 0.2067739045023918,
+      "eval_physics_loss": 1.3985861539840698,
+      "eval_physics_mean_token_accuracy": 0.7964418997764587,
+      "eval_physics_num_tokens": 231803509.0,
+      "eval_physics_runtime": 41.4682,
+      "eval_physics_samples_per_second": 12.057,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8100
+    },
+    {
+      "entropy": 0.05145827913656831,
+      "epoch": 12.975999999999999,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.202222222222222e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9905312418937683,
+      "num_tokens": 232095976.0,
+      "step": 8110
+    },
+    {
+      "entropy": 0.05109680832829326,
+      "epoch": 12.992,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903466764837503,
+      "num_tokens": 232382972.0,
+      "step": 8120
+    },
+    {
+      "entropy": 0.0512539794202894,
+      "epoch": 13.008,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.157777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906226065009832,
+      "num_tokens": 232672057.0,
+      "step": 8130
+    },
+    {
+      "entropy": 0.050676750240381804,
+      "epoch": 13.024,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.135555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990827102959156,
+      "num_tokens": 232958227.0,
+      "step": 8140
+    },
+    {
+      "entropy": 0.05172369135543704,
+      "epoch": 13.04,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.1133333333333335e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990977942943573,
+      "num_tokens": 233251931.0,
+      "step": 8150
+    },
+    {
+      "entropy": 0.052580399974249306,
+      "epoch": 13.056,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.091111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905320156365633,
+      "num_tokens": 233531328.0,
+      "step": 8160
+    },
+    {
+      "entropy": 0.05163608979200944,
+      "epoch": 13.072,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.0688888888888896e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990771321207285,
+      "num_tokens": 233816145.0,
+      "step": 8170
+    },
+    {
+      "entropy": 0.05149128795601428,
+      "epoch": 13.088,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.046666666666667e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.990656491741538,
+      "num_tokens": 234100261.0,
+      "step": 8180
+    },
+    {
+      "entropy": 0.0517811676254496,
+      "epoch": 13.104,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.024444444444445e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906980969011784,
+      "num_tokens": 234387753.0,
+      "step": 8190
+    },
+    {
+      "entropy": 0.051599945151247086,
+      "epoch": 13.12,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.002222222222222e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990650337934494,
+      "num_tokens": 234673838.0,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_math_entropy": 0.18022768545150758,
+      "eval_math_loss": 1.3355015516281128,
+      "eval_math_mean_token_accuracy": 0.8160842609405518,
+      "eval_math_num_tokens": 234673838.0,
+      "eval_math_runtime": 35.7873,
+      "eval_math_samples_per_second": 13.971,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_physics_entropy": 0.20687673270702361,
+      "eval_physics_loss": 1.401163935661316,
+      "eval_physics_mean_token_accuracy": 0.7964475116729737,
+      "eval_physics_num_tokens": 234673838.0,
+      "eval_physics_runtime": 41.4759,
+      "eval_physics_samples_per_second": 12.055,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8200
+    },
+    {
+      "entropy": 0.05104603223735467,
+      "epoch": 13.136,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908260378986598,
+      "num_tokens": 234966723.0,
+      "step": 8210
+    },
+    {
+      "entropy": 0.051453781058080496,
+      "epoch": 13.152,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.9577777777777785e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907778527587652,
+      "num_tokens": 235250894.0,
+      "step": 8220
+    },
+    {
+      "entropy": 0.052391895826440306,
+      "epoch": 13.168,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.935555555555556e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904574226588011,
+      "num_tokens": 235538789.0,
+      "step": 8230
+    },
+    {
+      "entropy": 0.05026549928588793,
+      "epoch": 13.184,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.913333333333334e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909756503999233,
+      "num_tokens": 235837152.0,
+      "step": 8240
+    },
+    {
+      "entropy": 0.053065391920972614,
+      "epoch": 13.2,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.891111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901943173259496,
+      "num_tokens": 236116206.0,
+      "step": 8250
+    },
+    {
+      "entropy": 0.051446165784727785,
+      "epoch": 13.216,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.868888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906851042062044,
+      "num_tokens": 236399981.0,
+      "step": 8260
+    },
+    {
+      "entropy": 0.0512137760524638,
+      "epoch": 13.232,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.8466666666666665e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906529590487481,
+      "num_tokens": 236687282.0,
+      "step": 8270
+    },
+    {
+      "entropy": 0.05217738675419241,
+      "epoch": 13.248,
+      "grad_norm": 0.5,
+      "learning_rate": 3.824444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905663672834635,
+      "num_tokens": 236970703.0,
+      "step": 8280
+    },
+    {
+      "entropy": 0.051728677074424925,
+      "epoch": 13.264,
+      "grad_norm": 0.52734375,
+      "learning_rate": 3.8022222222222226e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905138719826937,
+      "num_tokens": 237250995.0,
+      "step": 8290
+    },
+    {
+      "entropy": 0.05051842969842255,
+      "epoch": 13.28,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908084899187088,
+      "num_tokens": 237538556.0,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_math_entropy": 0.1792960543036461,
+      "eval_math_loss": 1.336340308189392,
+      "eval_math_mean_token_accuracy": 0.8159542636871338,
+      "eval_math_num_tokens": 237538556.0,
+      "eval_math_runtime": 35.8026,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_physics_entropy": 0.20614614540338516,
+      "eval_physics_loss": 1.4020493030548096,
+      "eval_physics_mean_token_accuracy": 0.796203773021698,
+      "eval_physics_num_tokens": 237538556.0,
+      "eval_physics_runtime": 41.473,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8300
+    },
+    {
+      "entropy": 0.05342790780123323,
+      "epoch": 13.296,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.757777777777778e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900773618370294,
+      "num_tokens": 237813501.0,
+      "step": 8310
+    },
+    {
+      "entropy": 0.05309537169523537,
+      "epoch": 13.312,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.7355555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903508022427558,
+      "num_tokens": 238094305.0,
+      "step": 8320
+    },
+    {
+      "entropy": 0.05170744662173092,
+      "epoch": 13.328,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.713333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905386842787266,
+      "num_tokens": 238374175.0,
+      "step": 8330
+    },
+    {
+      "entropy": 0.04992100042290985,
+      "epoch": 13.344,
+      "grad_norm": 0.41015625,
+      "learning_rate": 3.6911111111111115e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908522251993418,
+      "num_tokens": 238668031.0,
+      "step": 8340
+    },
+    {
+      "entropy": 0.05189812278840691,
+      "epoch": 13.36,
+      "grad_norm": 0.470703125,
+      "learning_rate": 3.668888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905831549316645,
+      "num_tokens": 238950369.0,
+      "step": 8350
+    },
+    {
+      "entropy": 0.05139565443387255,
+      "epoch": 13.376,
+      "grad_norm": 0.46875,
+      "learning_rate": 3.6466666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906327720731497,
+      "num_tokens": 239233027.0,
+      "step": 8360
+    },
+    {
+      "entropy": 0.05204238067381084,
+      "epoch": 13.392,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.624444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904446199536323,
+      "num_tokens": 239513675.0,
+      "step": 8370
+    },
+    {
+      "entropy": 0.051922239165287466,
+      "epoch": 13.408,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.6022222222222224e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9902938231825829,
+      "num_tokens": 239797199.0,
+      "step": 8380
+    },
+    {
+      "entropy": 0.05086992103606462,
+      "epoch": 13.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905360575765372,
+      "num_tokens": 240082617.0,
+      "step": 8390
+    },
+    {
+      "entropy": 0.052436576853506264,
+      "epoch": 13.44,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.5577777777777785e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903819940984249,
+      "num_tokens": 240365037.0,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_math_entropy": 0.17929287600517274,
+      "eval_math_loss": 1.3361586332321167,
+      "eval_math_mean_token_accuracy": 0.8159851579666137,
+      "eval_math_num_tokens": 240365037.0,
+      "eval_math_runtime": 35.8812,
+      "eval_math_samples_per_second": 13.935,
+      "eval_math_steps_per_second": 3.484,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_physics_entropy": 0.2061038536429405,
+      "eval_physics_loss": 1.4018964767456055,
+      "eval_physics_mean_token_accuracy": 0.7964254455566406,
+      "eval_physics_num_tokens": 240365037.0,
+      "eval_physics_runtime": 41.5825,
+      "eval_physics_samples_per_second": 12.024,
+      "eval_physics_steps_per_second": 3.006,
+      "step": 8400
+    },
+    {
+      "entropy": 0.052103328774683176,
+      "epoch": 13.456,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.535555555555556e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904427032917738,
+      "num_tokens": 240647742.0,
+      "step": 8410
+    },
+    {
+      "entropy": 0.0523280065273866,
+      "epoch": 13.472,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.5133333333333337e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904405385255813,
+      "num_tokens": 240926943.0,
+      "step": 8420
+    },
+    {
+      "entropy": 0.05302672478137538,
+      "epoch": 13.488,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4911111111111113e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903274435549975,
+      "num_tokens": 241206172.0,
+      "step": 8430
+    },
+    {
+      "entropy": 0.05202294115442783,
+      "epoch": 13.504,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4688888888888894e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905512318015098,
+      "num_tokens": 241489988.0,
+      "step": 8440
+    },
+    {
+      "entropy": 0.050564923824276777,
+      "epoch": 13.52,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.446666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907515056431293,
+      "num_tokens": 241781801.0,
+      "step": 8450
+    },
+    {
+      "entropy": 0.05287049149628729,
+      "epoch": 13.536,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.4244444444444446e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903701849281787,
+      "num_tokens": 242064444.0,
+      "step": 8460
+    },
+    {
+      "entropy": 0.05167013480095193,
+      "epoch": 13.552,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.4022222222222222e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904057901352644,
+      "num_tokens": 242354272.0,
+      "step": 8470
+    },
+    {
+      "entropy": 0.0516669807722792,
+      "epoch": 13.568,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9903885431587696,
+      "num_tokens": 242640475.0,
+      "step": 8480
+    },
+    {
+      "entropy": 0.05292565480340272,
+      "epoch": 13.584,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.3577777777777783e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901615660637617,
+      "num_tokens": 242920673.0,
+      "step": 8490
+    },
+    {
+      "entropy": 0.051914384588599205,
+      "epoch": 13.6,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990172453969717,
+      "num_tokens": 243198176.0,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_math_entropy": 0.17909148234128952,
+      "eval_math_loss": 1.3365639448165894,
+      "eval_math_mean_token_accuracy": 0.8161025867462158,
+      "eval_math_num_tokens": 243198176.0,
+      "eval_math_runtime": 35.7973,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_physics_entropy": 0.2059074219465256,
+      "eval_physics_loss": 1.4022636413574219,
+      "eval_physics_mean_token_accuracy": 0.7963426208496094,
+      "eval_physics_num_tokens": 243198176.0,
+      "eval_physics_runtime": 41.1599,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 8500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.843993565569015e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-8500/training_args.bin b/checkpoint-8500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-8500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-9000/chat_template.jinja b/checkpoint-9000/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-9000/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-9000/config.json b/checkpoint-9000/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-9000/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-9000/generation_config.json b/checkpoint-9000/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-9000/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-9000/model-00001-of-00004.safetensors b/checkpoint-9000/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-9000/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-9000/model-00002-of-00004.safetensors b/checkpoint-9000/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-9000/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-9000/model-00003-of-00004.safetensors b/checkpoint-9000/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..18cd1d7
--- /dev/null
+++ b/checkpoint-9000/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1faa17954c717de7fc38c5eb287843c368f7735b9b2637e64bca4846939be0e
+size 4962221464
diff --git a/checkpoint-9000/model-00004-of-00004.safetensors b/checkpoint-9000/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..f5086b7
--- /dev/null
+++ b/checkpoint-9000/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8508269115b4ea5ce61c95194d6f7d5f62b64e39f2a80b46bec87359f6a402c1
+size 3670322200
diff --git a/checkpoint-9000/model.safetensors.index.json b/checkpoint-9000/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-9000/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-9000/optimizer.pt b/checkpoint-9000/optimizer.pt
new file mode 100644
index 0000000..148514b
--- /dev/null
+++ b/checkpoint-9000/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:402207fa47551ef59653f7f1c8ad16093185358d03a307b51d6e259121252a4b
+size 7927918951
diff --git a/checkpoint-9000/rng_state.pth b/checkpoint-9000/rng_state.pth
new file mode 100644
index 0000000..48a34b3
--- /dev/null
+++ b/checkpoint-9000/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4c319ed98381004d56ec0f9a32b07389ddf647d317d2a3721107b3040de571f6
+size 14645
diff --git a/checkpoint-9000/scheduler.pt b/checkpoint-9000/scheduler.pt
new file mode 100644
index 0000000..3dc213a
--- /dev/null
+++ b/checkpoint-9000/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f813cb0e599c11b6d2eb3e4cb297b2fa03a67a86c14db7a8347f957f7cd30081
+size 1465
diff --git a/checkpoint-9000/special_tokens_map.json b/checkpoint-9000/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-9000/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-9000/tokenizer.json b/checkpoint-9000/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-9000/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-9000/tokenizer.model b/checkpoint-9000/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-9000/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-9000/tokenizer_config.json b/checkpoint-9000/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-9000/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-9000/trainer_state.json b/checkpoint-9000/trainer_state.json
new file mode 100644
index 0000000..73445a6
--- /dev/null
+++ b/checkpoint-9000/trainer_state.json
@@ -0,0 +1,11014 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 14.4,
+  "eval_steps": 100,
+  "global_step": 9000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    },
+    {
+      "entropy": 0.053174392925575374,
+      "epoch": 12.016,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.535555555555556e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905574269592762,
+      "num_tokens": 214920834.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.05121934600174427,
+      "epoch": 12.032,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.513333333333334e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907024279236794,
+      "num_tokens": 215207212.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.05282459647860378,
+      "epoch": 12.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.491111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904284842312336,
+      "num_tokens": 215491551.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.0523437969153747,
+      "epoch": 12.064,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.468888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907445590943098,
+      "num_tokens": 215773788.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.05109455803176388,
+      "epoch": 12.08,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990898734331131,
+      "num_tokens": 216063342.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.052485697087831795,
+      "epoch": 12.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.424444444444444e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905819371342659,
+      "num_tokens": 216346341.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.05154973501339555,
+      "epoch": 12.112,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.402222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906920839101077,
+      "num_tokens": 216631586.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.05027051754295826,
+      "epoch": 12.128,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908485073596239,
+      "num_tokens": 216921387.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.051271379156969485,
+      "epoch": 12.144,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.357777777777779e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907938674092293,
+      "num_tokens": 217208948.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.05280535111669451,
+      "epoch": 12.16,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904177527874708,
+      "num_tokens": 217488087.0,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_math_entropy": 0.17994125121831894,
+      "eval_math_loss": 1.3322224617004395,
+      "eval_math_mean_token_accuracy": 0.8160332980155944,
+      "eval_math_num_tokens": 217488087.0,
+      "eval_math_runtime": 35.8103,
+      "eval_math_samples_per_second": 13.962,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_physics_entropy": 0.20668217623233795,
+      "eval_physics_loss": 1.3987233638763428,
+      "eval_physics_mean_token_accuracy": 0.7963971982002258,
+      "eval_physics_num_tokens": 217488087.0,
+      "eval_physics_runtime": 41.4958,
+      "eval_physics_samples_per_second": 12.049,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7600
+    },
+    {
+      "entropy": 0.05196192401926965,
+      "epoch": 12.176,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.313333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906202860176563,
+      "num_tokens": 217770038.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.051902253658045086,
+      "epoch": 12.192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.2911111111111115e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9906028714030981,
+      "num_tokens": 218050486.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.052790193830151114,
+      "epoch": 12.208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.268888888888889e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904781151562929,
+      "num_tokens": 218331465.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.052298348920885473,
+      "epoch": 12.224,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.246666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906631808727979,
+      "num_tokens": 218618801.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.050798607314936815,
+      "epoch": 12.24,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9909202449023724,
+      "num_tokens": 218910596.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.052114113536663356,
+      "epoch": 12.256,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.202222222222223e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905635304749012,
+      "num_tokens": 219194842.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.05205410345224663,
+      "epoch": 12.272,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990602058544755,
+      "num_tokens": 219484076.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.049987882829736915,
+      "epoch": 12.288,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.157777777777778e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909869834780693,
+      "num_tokens": 219780222.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.05258604409173131,
+      "epoch": 12.304,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.135555555555556e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904522281140089,
+      "num_tokens": 220065272.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.05357218148419633,
+      "epoch": 12.32,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9902280792593956,
+      "num_tokens": 220344110.0,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_math_entropy": 0.1793700819015503,
+      "eval_math_loss": 1.3329575061798096,
+      "eval_math_mean_token_accuracy": 0.8161038937568664,
+      "eval_math_num_tokens": 220344110.0,
+      "eval_math_runtime": 35.8174,
+      "eval_math_samples_per_second": 13.96,
+      "eval_math_steps_per_second": 3.49,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_physics_entropy": 0.20628326386213303,
+      "eval_physics_loss": 1.3988101482391357,
+      "eval_physics_mean_token_accuracy": 0.7963225340843201,
+      "eval_physics_num_tokens": 220344110.0,
+      "eval_physics_runtime": 41.5047,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7700
+    },
+    {
+      "entropy": 0.05083432825049385,
+      "epoch": 12.336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.091111111111111e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906401328742505,
+      "num_tokens": 220632516.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.05188987164292484,
+      "epoch": 12.352,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.06888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905499298125505,
+      "num_tokens": 220918123.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.05209730467759073,
+      "epoch": 12.368,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.046666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905961699783802,
+      "num_tokens": 221206063.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.05238552400842309,
+      "epoch": 12.384,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.024444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904579646885395,
+      "num_tokens": 221488317.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.05181863873731345,
+      "epoch": 12.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905485171824694,
+      "num_tokens": 221776008.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.05192954689264297,
+      "epoch": 12.416,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904752410948277,
+      "num_tokens": 222057789.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.052794202556833626,
+      "epoch": 12.432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.957777777777778e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990259325504303,
+      "num_tokens": 222339915.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.052391084167174994,
+      "epoch": 12.448,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903523031622171,
+      "num_tokens": 222619687.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.05029154985677451,
+      "epoch": 12.464,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.9133333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906887255609036,
+      "num_tokens": 222913264.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.05082799810916185,
+      "epoch": 12.48,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906626105308532,
+      "num_tokens": 223202947.0,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_math_entropy": 0.180211255133152,
+      "eval_math_loss": 1.333121657371521,
+      "eval_math_mean_token_accuracy": 0.8160158872604371,
+      "eval_math_num_tokens": 223202947.0,
+      "eval_math_runtime": 35.8241,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_physics_entropy": 0.20699969559907913,
+      "eval_physics_loss": 1.3986775875091553,
+      "eval_physics_mean_token_accuracy": 0.7963728904724121,
+      "eval_physics_num_tokens": 223202947.0,
+      "eval_physics_runtime": 41.4724,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 7800
+    },
+    {
+      "entropy": 0.05148934966418892,
+      "epoch": 12.496,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.8688888888888895e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905773665755987,
+      "num_tokens": 223491040.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.05243352035759017,
+      "epoch": 12.512,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.846666666666667e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903561118990183,
+      "num_tokens": 223775825.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.05058162130881101,
+      "epoch": 12.528,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.824444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990830871462822,
+      "num_tokens": 224068092.0,
+      "step": 7830
+    },
+    {
+      "entropy": 0.051722967089153825,
+      "epoch": 12.544,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.802222222222222e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903257399797439,
+      "num_tokens": 224354624.0,
+      "step": 7840
+    },
+    {
+      "entropy": 0.05123099538031965,
+      "epoch": 12.56,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906783111393451,
+      "num_tokens": 224641974.0,
+      "step": 7850
+    },
+    {
+      "entropy": 0.05136660183779895,
+      "epoch": 12.576,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.7577777777777784e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906629983335733,
+      "num_tokens": 224932014.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.05145444863010198,
+      "epoch": 12.592,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.735555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9903172593563795,
+      "num_tokens": 225222434.0,
+      "step": 7870
+    },
+    {
+      "entropy": 0.05075638788985089,
+      "epoch": 12.608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.713333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906791742891073,
+      "num_tokens": 225515803.0,
+      "step": 7880
+    },
+    {
+      "entropy": 0.05076117676217109,
+      "epoch": 12.624,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.691111111111111e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906085334718228,
+      "num_tokens": 225810131.0,
+      "step": 7890
+    },
+    {
+      "entropy": 0.05141493146074936,
+      "epoch": 12.64,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904979523271322,
+      "num_tokens": 226100172.0,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_math_entropy": 0.17927622705698013,
+      "eval_math_loss": 1.332814335823059,
+      "eval_math_mean_token_accuracy": 0.8159863357543945,
+      "eval_math_num_tokens": 226100172.0,
+      "eval_math_runtime": 35.8058,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_physics_entropy": 0.20605486339330673,
+      "eval_physics_loss": 1.3984746932983398,
+      "eval_physics_mean_token_accuracy": 0.7964399604797363,
+      "eval_physics_num_tokens": 226100172.0,
+      "eval_physics_runtime": 41.5077,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7900
+    },
+    {
+      "entropy": 0.052684175770264116,
+      "epoch": 12.656,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.646666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902132380753755,
+      "num_tokens": 226379811.0,
+      "step": 7910
+    },
+    {
+      "entropy": 0.05348572533112019,
+      "epoch": 12.672,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.990258751064539,
+      "num_tokens": 226662085.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.05106104960432276,
+      "epoch": 12.688,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.602222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9902874868363142,
+      "num_tokens": 226950257.0,
+      "step": 7930
+    },
+    {
+      "entropy": 0.0518765551969409,
+      "epoch": 12.704,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905543696135283,
+      "num_tokens": 227238669.0,
+      "step": 7940
+    },
+    {
+      "entropy": 0.05253795627504587,
+      "epoch": 12.72,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990134946629405,
+      "num_tokens": 227518695.0,
+      "step": 7950
+    },
+    {
+      "entropy": 0.05207758563337848,
+      "epoch": 12.736,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.535555555555555e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990263070166111,
+      "num_tokens": 227806156.0,
+      "step": 7960
+    },
+    {
+      "entropy": 0.05179664236493409,
+      "epoch": 12.752,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.513333333333333e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904587935656309,
+      "num_tokens": 228091641.0,
+      "step": 7970
+    },
+    {
+      "entropy": 0.05190181992948055,
+      "epoch": 12.768,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.4911111111111115e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9901351545006036,
+      "num_tokens": 228376952.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.0528141722548753,
+      "epoch": 12.784,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.468888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901187218725681,
+      "num_tokens": 228657003.0,
+      "step": 7990
+    },
+    {
+      "entropy": 0.051391151722054926,
+      "epoch": 12.8,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.446666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9902074325829744,
+      "num_tokens": 228942799.0,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_math_entropy": 0.1799713587164879,
+      "eval_math_loss": 1.332999587059021,
+      "eval_math_mean_token_accuracy": 0.8159932594299316,
+      "eval_math_num_tokens": 228942799.0,
+      "eval_math_runtime": 35.8043,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_physics_entropy": 0.20658580523729325,
+      "eval_physics_loss": 1.3991401195526123,
+      "eval_physics_mean_token_accuracy": 0.7965000495910645,
+      "eval_physics_num_tokens": 228942799.0,
+      "eval_physics_runtime": 41.1662,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 8000
+    },
+    {
+      "entropy": 0.05222373401047662,
+      "epoch": 12.816,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.424444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902573928236962,
+      "num_tokens": 229230267.0,
+      "step": 8010
+    },
+    {
+      "entropy": 0.05221812457311899,
+      "epoch": 12.832,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.402222222222223e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902742177248001,
+      "num_tokens": 229513807.0,
+      "step": 8020
+    },
+    {
+      "entropy": 0.05192019739188254,
+      "epoch": 12.848,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990331218764186,
+      "num_tokens": 229800116.0,
+      "step": 8030
+    },
+    {
+      "entropy": 0.05204707682132721,
+      "epoch": 12.864,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904342792928219,
+      "num_tokens": 230086560.0,
+      "step": 8040
+    },
+    {
+      "entropy": 0.05261244847206399,
+      "epoch": 12.88,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.3355555555555565e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901342626661063,
+      "num_tokens": 230370705.0,
+      "step": 8050
+    },
+    {
+      "entropy": 0.050930758623871954,
+      "epoch": 12.896,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.313333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904414359480143,
+      "num_tokens": 230665090.0,
+      "step": 8060
+    },
+    {
+      "entropy": 0.0521988584427163,
+      "epoch": 12.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.291111111111112e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902524299919605,
+      "num_tokens": 230948774.0,
+      "step": 8070
+    },
+    {
+      "entropy": 0.05130416583269835,
+      "epoch": 12.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.268888888888889e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902212977409363,
+      "num_tokens": 231236122.0,
+      "step": 8080
+    },
+    {
+      "entropy": 0.05124916716013104,
+      "epoch": 12.943999999999999,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.246666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990312222391367,
+      "num_tokens": 231520994.0,
+      "step": 8090
+    },
+    {
+      "entropy": 0.05303966203937307,
+      "epoch": 12.96,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.2244444444444446e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9900147426873446,
+      "num_tokens": 231803509.0,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_math_entropy": 0.18036945784091948,
+      "eval_math_loss": 1.332650899887085,
+      "eval_math_mean_token_accuracy": 0.8161214051246644,
+      "eval_math_num_tokens": 231803509.0,
+      "eval_math_runtime": 35.8002,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_physics_entropy": 0.2067739045023918,
+      "eval_physics_loss": 1.3985861539840698,
+      "eval_physics_mean_token_accuracy": 0.7964418997764587,
+      "eval_physics_num_tokens": 231803509.0,
+      "eval_physics_runtime": 41.4682,
+      "eval_physics_samples_per_second": 12.057,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8100
+    },
+    {
+      "entropy": 0.05145827913656831,
+      "epoch": 12.975999999999999,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.202222222222222e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9905312418937683,
+      "num_tokens": 232095976.0,
+      "step": 8110
+    },
+    {
+      "entropy": 0.05109680832829326,
+      "epoch": 12.992,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903466764837503,
+      "num_tokens": 232382972.0,
+      "step": 8120
+    },
+    {
+      "entropy": 0.0512539794202894,
+      "epoch": 13.008,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.157777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906226065009832,
+      "num_tokens": 232672057.0,
+      "step": 8130
+    },
+    {
+      "entropy": 0.050676750240381804,
+      "epoch": 13.024,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.135555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990827102959156,
+      "num_tokens": 232958227.0,
+      "step": 8140
+    },
+    {
+      "entropy": 0.05172369135543704,
+      "epoch": 13.04,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.1133333333333335e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990977942943573,
+      "num_tokens": 233251931.0,
+      "step": 8150
+    },
+    {
+      "entropy": 0.052580399974249306,
+      "epoch": 13.056,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.091111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905320156365633,
+      "num_tokens": 233531328.0,
+      "step": 8160
+    },
+    {
+      "entropy": 0.05163608979200944,
+      "epoch": 13.072,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.0688888888888896e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990771321207285,
+      "num_tokens": 233816145.0,
+      "step": 8170
+    },
+    {
+      "entropy": 0.05149128795601428,
+      "epoch": 13.088,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.046666666666667e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.990656491741538,
+      "num_tokens": 234100261.0,
+      "step": 8180
+    },
+    {
+      "entropy": 0.0517811676254496,
+      "epoch": 13.104,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.024444444444445e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906980969011784,
+      "num_tokens": 234387753.0,
+      "step": 8190
+    },
+    {
+      "entropy": 0.051599945151247086,
+      "epoch": 13.12,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.002222222222222e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990650337934494,
+      "num_tokens": 234673838.0,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_math_entropy": 0.18022768545150758,
+      "eval_math_loss": 1.3355015516281128,
+      "eval_math_mean_token_accuracy": 0.8160842609405518,
+      "eval_math_num_tokens": 234673838.0,
+      "eval_math_runtime": 35.7873,
+      "eval_math_samples_per_second": 13.971,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_physics_entropy": 0.20687673270702361,
+      "eval_physics_loss": 1.401163935661316,
+      "eval_physics_mean_token_accuracy": 0.7964475116729737,
+      "eval_physics_num_tokens": 234673838.0,
+      "eval_physics_runtime": 41.4759,
+      "eval_physics_samples_per_second": 12.055,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8200
+    },
+    {
+      "entropy": 0.05104603223735467,
+      "epoch": 13.136,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908260378986598,
+      "num_tokens": 234966723.0,
+      "step": 8210
+    },
+    {
+      "entropy": 0.051453781058080496,
+      "epoch": 13.152,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.9577777777777785e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907778527587652,
+      "num_tokens": 235250894.0,
+      "step": 8220
+    },
+    {
+      "entropy": 0.052391895826440306,
+      "epoch": 13.168,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.935555555555556e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904574226588011,
+      "num_tokens": 235538789.0,
+      "step": 8230
+    },
+    {
+      "entropy": 0.05026549928588793,
+      "epoch": 13.184,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.913333333333334e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909756503999233,
+      "num_tokens": 235837152.0,
+      "step": 8240
+    },
+    {
+      "entropy": 0.053065391920972614,
+      "epoch": 13.2,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.891111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901943173259496,
+      "num_tokens": 236116206.0,
+      "step": 8250
+    },
+    {
+      "entropy": 0.051446165784727785,
+      "epoch": 13.216,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.868888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906851042062044,
+      "num_tokens": 236399981.0,
+      "step": 8260
+    },
+    {
+      "entropy": 0.0512137760524638,
+      "epoch": 13.232,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.8466666666666665e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906529590487481,
+      "num_tokens": 236687282.0,
+      "step": 8270
+    },
+    {
+      "entropy": 0.05217738675419241,
+      "epoch": 13.248,
+      "grad_norm": 0.5,
+      "learning_rate": 3.824444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905663672834635,
+      "num_tokens": 236970703.0,
+      "step": 8280
+    },
+    {
+      "entropy": 0.051728677074424925,
+      "epoch": 13.264,
+      "grad_norm": 0.52734375,
+      "learning_rate": 3.8022222222222226e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905138719826937,
+      "num_tokens": 237250995.0,
+      "step": 8290
+    },
+    {
+      "entropy": 0.05051842969842255,
+      "epoch": 13.28,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908084899187088,
+      "num_tokens": 237538556.0,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_math_entropy": 0.1792960543036461,
+      "eval_math_loss": 1.336340308189392,
+      "eval_math_mean_token_accuracy": 0.8159542636871338,
+      "eval_math_num_tokens": 237538556.0,
+      "eval_math_runtime": 35.8026,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_physics_entropy": 0.20614614540338516,
+      "eval_physics_loss": 1.4020493030548096,
+      "eval_physics_mean_token_accuracy": 0.796203773021698,
+      "eval_physics_num_tokens": 237538556.0,
+      "eval_physics_runtime": 41.473,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8300
+    },
+    {
+      "entropy": 0.05342790780123323,
+      "epoch": 13.296,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.757777777777778e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900773618370294,
+      "num_tokens": 237813501.0,
+      "step": 8310
+    },
+    {
+      "entropy": 0.05309537169523537,
+      "epoch": 13.312,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.7355555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903508022427558,
+      "num_tokens": 238094305.0,
+      "step": 8320
+    },
+    {
+      "entropy": 0.05170744662173092,
+      "epoch": 13.328,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.713333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905386842787266,
+      "num_tokens": 238374175.0,
+      "step": 8330
+    },
+    {
+      "entropy": 0.04992100042290985,
+      "epoch": 13.344,
+      "grad_norm": 0.41015625,
+      "learning_rate": 3.6911111111111115e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908522251993418,
+      "num_tokens": 238668031.0,
+      "step": 8340
+    },
+    {
+      "entropy": 0.05189812278840691,
+      "epoch": 13.36,
+      "grad_norm": 0.470703125,
+      "learning_rate": 3.668888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905831549316645,
+      "num_tokens": 238950369.0,
+      "step": 8350
+    },
+    {
+      "entropy": 0.05139565443387255,
+      "epoch": 13.376,
+      "grad_norm": 0.46875,
+      "learning_rate": 3.6466666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906327720731497,
+      "num_tokens": 239233027.0,
+      "step": 8360
+    },
+    {
+      "entropy": 0.05204238067381084,
+      "epoch": 13.392,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.624444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904446199536323,
+      "num_tokens": 239513675.0,
+      "step": 8370
+    },
+    {
+      "entropy": 0.051922239165287466,
+      "epoch": 13.408,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.6022222222222224e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9902938231825829,
+      "num_tokens": 239797199.0,
+      "step": 8380
+    },
+    {
+      "entropy": 0.05086992103606462,
+      "epoch": 13.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905360575765372,
+      "num_tokens": 240082617.0,
+      "step": 8390
+    },
+    {
+      "entropy": 0.052436576853506264,
+      "epoch": 13.44,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.5577777777777785e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903819940984249,
+      "num_tokens": 240365037.0,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_math_entropy": 0.17929287600517274,
+      "eval_math_loss": 1.3361586332321167,
+      "eval_math_mean_token_accuracy": 0.8159851579666137,
+      "eval_math_num_tokens": 240365037.0,
+      "eval_math_runtime": 35.8812,
+      "eval_math_samples_per_second": 13.935,
+      "eval_math_steps_per_second": 3.484,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_physics_entropy": 0.2061038536429405,
+      "eval_physics_loss": 1.4018964767456055,
+      "eval_physics_mean_token_accuracy": 0.7964254455566406,
+      "eval_physics_num_tokens": 240365037.0,
+      "eval_physics_runtime": 41.5825,
+      "eval_physics_samples_per_second": 12.024,
+      "eval_physics_steps_per_second": 3.006,
+      "step": 8400
+    },
+    {
+      "entropy": 0.052103328774683176,
+      "epoch": 13.456,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.535555555555556e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904427032917738,
+      "num_tokens": 240647742.0,
+      "step": 8410
+    },
+    {
+      "entropy": 0.0523280065273866,
+      "epoch": 13.472,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.5133333333333337e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904405385255813,
+      "num_tokens": 240926943.0,
+      "step": 8420
+    },
+    {
+      "entropy": 0.05302672478137538,
+      "epoch": 13.488,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4911111111111113e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903274435549975,
+      "num_tokens": 241206172.0,
+      "step": 8430
+    },
+    {
+      "entropy": 0.05202294115442783,
+      "epoch": 13.504,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4688888888888894e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905512318015098,
+      "num_tokens": 241489988.0,
+      "step": 8440
+    },
+    {
+      "entropy": 0.050564923824276777,
+      "epoch": 13.52,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.446666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907515056431293,
+      "num_tokens": 241781801.0,
+      "step": 8450
+    },
+    {
+      "entropy": 0.05287049149628729,
+      "epoch": 13.536,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.4244444444444446e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903701849281787,
+      "num_tokens": 242064444.0,
+      "step": 8460
+    },
+    {
+      "entropy": 0.05167013480095193,
+      "epoch": 13.552,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.4022222222222222e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904057901352644,
+      "num_tokens": 242354272.0,
+      "step": 8470
+    },
+    {
+      "entropy": 0.0516669807722792,
+      "epoch": 13.568,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9903885431587696,
+      "num_tokens": 242640475.0,
+      "step": 8480
+    },
+    {
+      "entropy": 0.05292565480340272,
+      "epoch": 13.584,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.3577777777777783e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901615660637617,
+      "num_tokens": 242920673.0,
+      "step": 8490
+    },
+    {
+      "entropy": 0.051914384588599205,
+      "epoch": 13.6,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990172453969717,
+      "num_tokens": 243198176.0,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_math_entropy": 0.17909148234128952,
+      "eval_math_loss": 1.3365639448165894,
+      "eval_math_mean_token_accuracy": 0.8161025867462158,
+      "eval_math_num_tokens": 243198176.0,
+      "eval_math_runtime": 35.7973,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_physics_entropy": 0.2059074219465256,
+      "eval_physics_loss": 1.4022636413574219,
+      "eval_physics_mean_token_accuracy": 0.7963426208496094,
+      "eval_physics_num_tokens": 243198176.0,
+      "eval_physics_runtime": 41.1599,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 8500
+    },
+    {
+      "entropy": 0.05167351650306955,
+      "epoch": 13.616,
+      "grad_norm": 0.478515625,
+      "learning_rate": 3.3133333333333335e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904801592230796,
+      "num_tokens": 243483881.0,
+      "step": 8510
+    },
+    {
+      "entropy": 0.053366528800688685,
+      "epoch": 13.632,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.2911111111111116e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9902401711791754,
+      "num_tokens": 243764128.0,
+      "step": 8520
+    },
+    {
+      "entropy": 0.05087098733056337,
+      "epoch": 13.648,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904976122081279,
+      "num_tokens": 244051585.0,
+      "step": 8530
+    },
+    {
+      "entropy": 0.05087883236119524,
+      "epoch": 13.664,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.2466666666666668e-06,
+      "loss": 0.0302,
+      "mean_token_accuracy": 0.9905655104666948,
+      "num_tokens": 244349004.0,
+      "step": 8540
+    },
+    {
+      "entropy": 0.04904007929144427,
+      "epoch": 13.68,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.2244444444444444e-06,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.990977830812335,
+      "num_tokens": 244649637.0,
+      "step": 8550
+    },
+    {
+      "entropy": 0.0504306229762733,
+      "epoch": 13.696,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.202222222222223e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907431546598673,
+      "num_tokens": 244940362.0,
+      "step": 8560
+    },
+    {
+      "entropy": 0.05084956689970568,
+      "epoch": 13.712,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904598340392112,
+      "num_tokens": 245228064.0,
+      "step": 8570
+    },
+    {
+      "entropy": 0.05026600864948705,
+      "epoch": 13.728,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.157777777777778e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9905784897506237,
+      "num_tokens": 245516918.0,
+      "step": 8580
+    },
+    {
+      "entropy": 0.050801993359345946,
+      "epoch": 13.744,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.1355555555555557e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907515291124582,
+      "num_tokens": 245807811.0,
+      "step": 8590
+    },
+    {
+      "entropy": 0.05082948401104659,
+      "epoch": 13.76,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.1133333333333337e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9906744692474604,
+      "num_tokens": 246099779.0,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_math_entropy": 0.17962482076883315,
+      "eval_math_loss": 1.3373396396636963,
+      "eval_math_mean_token_accuracy": 0.816063407421112,
+      "eval_math_num_tokens": 246099779.0,
+      "eval_math_runtime": 35.7911,
+      "eval_math_samples_per_second": 13.97,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_physics_entropy": 0.20641914105415343,
+      "eval_physics_loss": 1.4027777910232544,
+      "eval_physics_mean_token_accuracy": 0.796207398891449,
+      "eval_physics_num_tokens": 246099779.0,
+      "eval_physics_runtime": 41.482,
+      "eval_physics_samples_per_second": 12.053,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 8600
+    },
+    {
+      "entropy": 0.05176880229264498,
+      "epoch": 13.776,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0911111111111114e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904642947018146,
+      "num_tokens": 246383665.0,
+      "step": 8610
+    },
+    {
+      "entropy": 0.05162258397322148,
+      "epoch": 13.792,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.068888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904142130166292,
+      "num_tokens": 246668395.0,
+      "step": 8620
+    },
+    {
+      "entropy": 0.0508142440347001,
+      "epoch": 13.808,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.0466666666666666e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907273758202791,
+      "num_tokens": 246960911.0,
+      "step": 8630
+    },
+    {
+      "entropy": 0.04985311826458201,
+      "epoch": 13.824,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.024444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908174972981214,
+      "num_tokens": 247253562.0,
+      "step": 8640
+    },
+    {
+      "entropy": 0.052733641443774106,
+      "epoch": 13.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0022222222222227e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9900983639061451,
+      "num_tokens": 247529533.0,
+      "step": 8650
+    },
+    {
+      "entropy": 0.05027402397245169,
+      "epoch": 13.856,
+      "grad_norm": 0.45703125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990868479758501,
+      "num_tokens": 247819702.0,
+      "step": 8660
+    },
+    {
+      "entropy": 0.050137166888453065,
+      "epoch": 13.872,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.957777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907190252095461,
+      "num_tokens": 248113990.0,
+      "step": 8670
+    },
+    {
+      "entropy": 0.0517217613523826,
+      "epoch": 13.888,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.935555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902944087982177,
+      "num_tokens": 248397281.0,
+      "step": 8680
+    },
+    {
+      "entropy": 0.05074853647965938,
+      "epoch": 13.904,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.9133333333333335e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9904990620911122,
+      "num_tokens": 248689496.0,
+      "step": 8690
+    },
+    {
+      "entropy": 0.05088953715749085,
+      "epoch": 13.92,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.891111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905771657824516,
+      "num_tokens": 248975741.0,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_math_entropy": 0.17970398700237275,
+      "eval_math_loss": 1.336698293685913,
+      "eval_math_mean_token_accuracy": 0.8159835457801818,
+      "eval_math_num_tokens": 248975741.0,
+      "eval_math_runtime": 35.7809,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_physics_entropy": 0.20640488415956498,
+      "eval_physics_loss": 1.4024730920791626,
+      "eval_physics_mean_token_accuracy": 0.7963580441474915,
+      "eval_physics_num_tokens": 248975741.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8700
+    },
+    {
+      "entropy": 0.04994275859789923,
+      "epoch": 13.936,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.868888888888889e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9907119512557984,
+      "num_tokens": 249270221.0,
+      "step": 8710
+    },
+    {
+      "entropy": 0.04963390504708513,
+      "epoch": 13.952,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.8466666666666672e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9907703123986721,
+      "num_tokens": 249563652.0,
+      "step": 8720
+    },
+    {
+      "entropy": 0.05193966852966696,
+      "epoch": 13.968,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.824444444444445e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902710720896721,
+      "num_tokens": 249847652.0,
+      "step": 8730
+    },
+    {
+      "entropy": 0.05370221418561414,
+      "epoch": 13.984,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.8022222222222225e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9897177372127771,
+      "num_tokens": 250120462.0,
+      "step": 8740
+    },
+    {
+      "entropy": 0.05114520986098796,
+      "epoch": 14.0,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9904374986886978,
+      "num_tokens": 250409936.0,
+      "step": 8750
+    },
+    {
+      "entropy": 0.05091056008823216,
+      "epoch": 14.016,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.757777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908093482255935,
+      "num_tokens": 250696959.0,
+      "step": 8760
+    },
+    {
+      "entropy": 0.052239940082654354,
+      "epoch": 14.032,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.7355555555555557e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906717732548713,
+      "num_tokens": 250980823.0,
+      "step": 8770
+    },
+    {
+      "entropy": 0.051255598349962385,
+      "epoch": 14.048,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.7133333333333333e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990817604586482,
+      "num_tokens": 251268212.0,
+      "step": 8780
+    },
+    {
+      "entropy": 0.05210095479851588,
+      "epoch": 14.064,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.6911111111111114e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904542353004218,
+      "num_tokens": 251553952.0,
+      "step": 8790
+    },
+    {
+      "entropy": 0.05244319257326424,
+      "epoch": 14.08,
+      "grad_norm": 0.458984375,
+      "learning_rate": 2.6688888888888894e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904410917311907,
+      "num_tokens": 251838168.0,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_math_entropy": 0.17896863031387328,
+      "eval_math_loss": 1.3373428583145142,
+      "eval_math_mean_token_accuracy": 0.8160420455932618,
+      "eval_math_num_tokens": 251838168.0,
+      "eval_math_runtime": 35.776,
+      "eval_math_samples_per_second": 13.976,
+      "eval_math_steps_per_second": 3.494,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_physics_entropy": 0.20583795410394667,
+      "eval_physics_loss": 1.4028042554855347,
+      "eval_physics_mean_token_accuracy": 0.7963007631301879,
+      "eval_physics_num_tokens": 251838168.0,
+      "eval_physics_runtime": 41.4723,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8800
+    },
+    {
+      "entropy": 0.05141720774117857,
+      "epoch": 14.096,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.646666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906542908400298,
+      "num_tokens": 252121559.0,
+      "step": 8810
+    },
+    {
+      "entropy": 0.050573783030267806,
+      "epoch": 14.112,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.6244444444444446e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9908661898225546,
+      "num_tokens": 252413324.0,
+      "step": 8820
+    },
+    {
+      "entropy": 0.05072221851442009,
+      "epoch": 14.128,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.6022222222222227e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908107925206423,
+      "num_tokens": 252700652.0,
+      "step": 8830
+    },
+    {
+      "entropy": 0.04975284937536344,
+      "epoch": 14.144,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910169199109078,
+      "num_tokens": 252992593.0,
+      "step": 8840
+    },
+    {
+      "entropy": 0.049413899821229276,
+      "epoch": 14.16,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.557777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910381846129894,
+      "num_tokens": 253284586.0,
+      "step": 8850
+    },
+    {
+      "entropy": 0.04987312061712146,
+      "epoch": 14.176,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.5355555555555555e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.9910611238330602,
+      "num_tokens": 253580393.0,
+      "step": 8860
+    },
+    {
+      "entropy": 0.05002628688234836,
+      "epoch": 14.192,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.5133333333333336e-06,
+      "loss": 0.0303,
+      "mean_token_accuracy": 0.9909944098442793,
+      "num_tokens": 253873719.0,
+      "step": 8870
+    },
+    {
+      "entropy": 0.051363390649203214,
+      "epoch": 14.208,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.491111111111111e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.990709126740694,
+      "num_tokens": 254163170.0,
+      "step": 8880
+    },
+    {
+      "entropy": 0.05138509668176994,
+      "epoch": 14.224,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905464217066765,
+      "num_tokens": 254445476.0,
+      "step": 8890
+    },
+    {
+      "entropy": 0.051887833909131584,
+      "epoch": 14.24,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.446666666666667e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906827192753553,
+      "num_tokens": 254731054.0,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_math_entropy": 0.17892871648073197,
+      "eval_math_loss": 1.3377963304519653,
+      "eval_math_mean_token_accuracy": 0.8161516799926758,
+      "eval_math_num_tokens": 254731054.0,
+      "eval_math_runtime": 35.7685,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_physics_entropy": 0.20582070446014405,
+      "eval_physics_loss": 1.4032886028289795,
+      "eval_physics_mean_token_accuracy": 0.7961821508407593,
+      "eval_physics_num_tokens": 254731054.0,
+      "eval_physics_runtime": 41.4536,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8900
+    },
+    {
+      "entropy": 0.05053608596790582,
+      "epoch": 14.256,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.4244444444444444e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908701453357935,
+      "num_tokens": 255017263.0,
+      "step": 8910
+    },
+    {
+      "entropy": 0.04994903282495215,
+      "epoch": 14.272,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.4022222222222225e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907931432127952,
+      "num_tokens": 255310045.0,
+      "step": 8920
+    },
+    {
+      "entropy": 0.0498261819826439,
+      "epoch": 14.288,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.38e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908365543931723,
+      "num_tokens": 255601380.0,
+      "step": 8930
+    },
+    {
+      "entropy": 0.05181859011063352,
+      "epoch": 14.304,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.357777777777778e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906132340431213,
+      "num_tokens": 255886699.0,
+      "step": 8940
+    },
+    {
+      "entropy": 0.0507721847621724,
+      "epoch": 14.32,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.3355555555555557e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9908549074083567,
+      "num_tokens": 256174927.0,
+      "step": 8950
+    },
+    {
+      "entropy": 0.05199292111210525,
+      "epoch": 14.336,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.3133333333333333e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990635683387518,
+      "num_tokens": 256456044.0,
+      "step": 8960
+    },
+    {
+      "entropy": 0.051074851502198725,
+      "epoch": 14.352,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.2911111111111114e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907376989722252,
+      "num_tokens": 256737497.0,
+      "step": 8970
+    },
+    {
+      "entropy": 0.050048362242523584,
+      "epoch": 14.368,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.268888888888889e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909071121364832,
+      "num_tokens": 257032721.0,
+      "step": 8980
+    },
+    {
+      "entropy": 0.052360807289369406,
+      "epoch": 14.384,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.2466666666666666e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905145078897476,
+      "num_tokens": 257311728.0,
+      "step": 8990
+    },
+    {
+      "entropy": 0.051360888360068205,
+      "epoch": 14.4,
+      "grad_norm": 0.4921875,
+      "learning_rate": 2.2244444444444447e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906007140874863,
+      "num_tokens": 257598724.0,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_math_entropy": 0.17880775159597398,
+      "eval_math_loss": 1.3379243612289429,
+      "eval_math_mean_token_accuracy": 0.8160511145591736,
+      "eval_math_num_tokens": 257598724.0,
+      "eval_math_runtime": 35.7772,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_physics_entropy": 0.20577423387765884,
+      "eval_physics_loss": 1.4032671451568604,
+      "eval_physics_mean_token_accuracy": 0.7963672475814819,
+      "eval_physics_num_tokens": 257598724.0,
+      "eval_physics_runtime": 41.5292,
+      "eval_physics_samples_per_second": 12.04,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 9000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.953118559386793e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-9000/training_args.bin b/checkpoint-9000/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-9000/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/checkpoint-9500/chat_template.jinja b/checkpoint-9500/chat_template.jinja
new file mode 100644
index 0000000..923ec25
--- /dev/null
+++ b/checkpoint-9500/chat_template.jinja
@@ -0,0 +1,4 @@
+{{ bos_token }}{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in messages %}{% if (message['role'] == 'user') != (loop.index0 % 2 == 0) %}{{ raise_exception('Conversation roles must alternate user/assistant/user/assistant/...') }}{% endif %}{% if (message['role'] == 'assistant') %}{% set role = 'model' %}{% else %}{% set role = message['role'] %}{% endif %}{{ '<start_of_turn>' + role + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}
\ No newline at end of file
diff --git a/checkpoint-9500/config.json b/checkpoint-9500/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/checkpoint-9500/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/checkpoint-9500/generation_config.json b/checkpoint-9500/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/checkpoint-9500/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/checkpoint-9500/model-00001-of-00004.safetensors b/checkpoint-9500/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/checkpoint-9500/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/checkpoint-9500/model-00002-of-00004.safetensors b/checkpoint-9500/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/checkpoint-9500/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/checkpoint-9500/model-00003-of-00004.safetensors b/checkpoint-9500/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..5efbf82
--- /dev/null
+++ b/checkpoint-9500/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f570418e8f202750eb24fad9329f4dd2d433313749e92b70b91e6b3f45e32e73
+size 4962221464
diff --git a/checkpoint-9500/model-00004-of-00004.safetensors b/checkpoint-9500/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..799b960
--- /dev/null
+++ b/checkpoint-9500/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ca5fee74a4f5aa03b6a495c7bdabf04180429299e8e7b15ca9e0c0b52e7e30e5
+size 3670322200
diff --git a/checkpoint-9500/model.safetensors.index.json b/checkpoint-9500/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/checkpoint-9500/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/checkpoint-9500/optimizer.pt b/checkpoint-9500/optimizer.pt
new file mode 100644
index 0000000..70de298
--- /dev/null
+++ b/checkpoint-9500/optimizer.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4306a91e5f3adc5fde3541b596a9ef5113033fabdbdb51852f7f3ea75269c68f
+size 7927918951
diff --git a/checkpoint-9500/rng_state.pth b/checkpoint-9500/rng_state.pth
new file mode 100644
index 0000000..d251a8b
--- /dev/null
+++ b/checkpoint-9500/rng_state.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a150b40e41b9473f4db252a72111069983fe4e32294e9cd18eda50b9fdc5e59a
+size 14645
diff --git a/checkpoint-9500/scheduler.pt b/checkpoint-9500/scheduler.pt
new file mode 100644
index 0000000..03f6fdf
--- /dev/null
+++ b/checkpoint-9500/scheduler.pt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:335e808e17d3b6faec21a0b84f5af0516d43f68cee4d7bbe101d896d991816d5
+size 1465
diff --git a/checkpoint-9500/special_tokens_map.json b/checkpoint-9500/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/checkpoint-9500/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/checkpoint-9500/tokenizer.json b/checkpoint-9500/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/checkpoint-9500/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/checkpoint-9500/tokenizer.model b/checkpoint-9500/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/checkpoint-9500/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/checkpoint-9500/tokenizer_config.json b/checkpoint-9500/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/checkpoint-9500/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}
diff --git a/checkpoint-9500/trainer_state.json b/checkpoint-9500/trainer_state.json
new file mode 100644
index 0000000..bb79f45
--- /dev/null
+++ b/checkpoint-9500/trainer_state.json
@@ -0,0 +1,11624 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 15.2,
+  "eval_steps": 100,
+  "global_step": 9500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "entropy": 0.6421742392703891,
+      "epoch": 0.016,
+      "grad_norm": 4.4375,
+      "learning_rate": 1.8e-07,
+      "loss": 0.953,
+      "mean_token_accuracy": 0.792544937506318,
+      "num_tokens": 282849.0,
+      "step": 10
+    },
+    {
+      "entropy": 0.6719903893768787,
+      "epoch": 0.032,
+      "grad_norm": 4.25,
+      "learning_rate": 3.8e-07,
+      "loss": 0.9758,
+      "mean_token_accuracy": 0.7857426963746548,
+      "num_tokens": 561726.0,
+      "step": 20
+    },
+    {
+      "entropy": 0.6603780068457127,
+      "epoch": 0.048,
+      "grad_norm": 4.40625,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 0.9582,
+      "mean_token_accuracy": 0.7894285101443529,
+      "num_tokens": 842092.0,
+      "step": 30
+    },
+    {
+      "entropy": 0.6775440879166126,
+      "epoch": 0.064,
+      "grad_norm": 3.890625,
+      "learning_rate": 7.8e-07,
+      "loss": 0.9776,
+      "mean_token_accuracy": 0.785004175081849,
+      "num_tokens": 1133349.0,
+      "step": 40
+    },
+    {
+      "entropy": 0.6926743468269706,
+      "epoch": 0.08,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 0.9497,
+      "mean_token_accuracy": 0.7896979171782732,
+      "num_tokens": 1422340.0,
+      "step": 50
+    },
+    {
+      "entropy": 0.7043748194351792,
+      "epoch": 0.096,
+      "grad_norm": 3.484375,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.9217,
+      "mean_token_accuracy": 0.7949878334999084,
+      "num_tokens": 1711821.0,
+      "step": 60
+    },
+    {
+      "entropy": 0.7551965553313493,
+      "epoch": 0.112,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.931,
+      "mean_token_accuracy": 0.7912098452448845,
+      "num_tokens": 1991855.0,
+      "step": 70
+    },
+    {
+      "entropy": 0.7913430579006672,
+      "epoch": 0.128,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.9172,
+      "mean_token_accuracy": 0.7935858864337206,
+      "num_tokens": 2288720.0,
+      "step": 80
+    },
+    {
+      "entropy": 0.7881425807252527,
+      "epoch": 0.144,
+      "grad_norm": 2.125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.917,
+      "mean_token_accuracy": 0.7932404045015573,
+      "num_tokens": 2564352.0,
+      "step": 90
+    },
+    {
+      "entropy": 0.7627748234197498,
+      "epoch": 0.16,
+      "grad_norm": 2.125,
+      "learning_rate": 1.98e-06,
+      "loss": 0.9014,
+      "mean_token_accuracy": 0.7970093633979559,
+      "num_tokens": 2841876.0,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_math_entropy": 0.7972113332748413,
+      "eval_math_loss": 0.9417253136634827,
+      "eval_math_mean_token_accuracy": 0.7907100429534912,
+      "eval_math_num_tokens": 2841876.0,
+      "eval_math_runtime": 35.8728,
+      "eval_math_samples_per_second": 13.938,
+      "eval_math_steps_per_second": 3.485,
+      "step": 100
+    },
+    {
+      "epoch": 0.16,
+      "eval_physics_entropy": 0.7183168392181396,
+      "eval_physics_loss": 0.7405859231948853,
+      "eval_physics_mean_token_accuracy": 0.822186062335968,
+      "eval_physics_num_tokens": 2841876.0,
+      "eval_physics_runtime": 41.4814,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 100
+    },
+    {
+      "entropy": 0.7770459342747926,
+      "epoch": 0.176,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.8952,
+      "mean_token_accuracy": 0.7969978783279658,
+      "num_tokens": 3129237.0,
+      "step": 110
+    },
+    {
+      "entropy": 0.7958783155307174,
+      "epoch": 0.192,
+      "grad_norm": 2.3125,
+      "learning_rate": 2.38e-06,
+      "loss": 0.9119,
+      "mean_token_accuracy": 0.793396720290184,
+      "num_tokens": 3414875.0,
+      "step": 120
+    },
+    {
+      "entropy": 0.7911858700215817,
+      "epoch": 0.208,
+      "grad_norm": 2.4375,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.8979,
+      "mean_token_accuracy": 0.7956417176872492,
+      "num_tokens": 3695623.0,
+      "step": 130
+    },
+    {
+      "entropy": 0.7871318949386478,
+      "epoch": 0.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.8882,
+      "mean_token_accuracy": 0.7958744909614325,
+      "num_tokens": 3977666.0,
+      "step": 140
+    },
+    {
+      "entropy": 0.7629930060356855,
+      "epoch": 0.24,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.8654,
+      "mean_token_accuracy": 0.8001766365021468,
+      "num_tokens": 4269287.0,
+      "step": 150
+    },
+    {
+      "entropy": 0.7563591374084353,
+      "epoch": 0.256,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.846,
+      "mean_token_accuracy": 0.8012917254120111,
+      "num_tokens": 4558632.0,
+      "step": 160
+    },
+    {
+      "entropy": 0.7706880155950785,
+      "epoch": 0.272,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.8589,
+      "mean_token_accuracy": 0.8001386649906636,
+      "num_tokens": 4839270.0,
+      "step": 170
+    },
+    {
+      "entropy": 0.7551738245412707,
+      "epoch": 0.288,
+      "grad_norm": 1.609375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.8468,
+      "mean_token_accuracy": 0.8015156831592322,
+      "num_tokens": 5125447.0,
+      "step": 180
+    },
+    {
+      "entropy": 0.7806578669697046,
+      "epoch": 0.304,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.8405,
+      "mean_token_accuracy": 0.7998461078852415,
+      "num_tokens": 5415925.0,
+      "step": 190
+    },
+    {
+      "entropy": 0.7316340124234557,
+      "epoch": 0.32,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.7942,
+      "mean_token_accuracy": 0.8076768793165684,
+      "num_tokens": 5710178.0,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_math_entropy": 0.7726770024299622,
+      "eval_math_loss": 0.84871506690979,
+      "eval_math_mean_token_accuracy": 0.7979933972358704,
+      "eval_math_num_tokens": 5710178.0,
+      "eval_math_runtime": 35.765,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 200
+    },
+    {
+      "epoch": 0.32,
+      "eval_physics_entropy": 0.6965276994705201,
+      "eval_physics_loss": 0.7036627531051636,
+      "eval_physics_mean_token_accuracy": 0.820941798210144,
+      "eval_physics_num_tokens": 5710178.0,
+      "eval_physics_runtime": 41.4597,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 200
+    },
+    {
+      "entropy": 0.7737893080338836,
+      "epoch": 0.336,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.8126,
+      "mean_token_accuracy": 0.800426610186696,
+      "num_tokens": 5993843.0,
+      "step": 210
+    },
+    {
+      "entropy": 0.7482354704290628,
+      "epoch": 0.352,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.7874,
+      "mean_token_accuracy": 0.8061643201857805,
+      "num_tokens": 6279137.0,
+      "step": 220
+    },
+    {
+      "entropy": 0.7608217550441623,
+      "epoch": 0.368,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.58e-06,
+      "loss": 0.7742,
+      "mean_token_accuracy": 0.803000221401453,
+      "num_tokens": 6562387.0,
+      "step": 230
+    },
+    {
+      "entropy": 0.748550821095705,
+      "epoch": 0.384,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.7565,
+      "mean_token_accuracy": 0.8081529751420021,
+      "num_tokens": 6851520.0,
+      "step": 240
+    },
+    {
+      "entropy": 0.7808488950133323,
+      "epoch": 0.4,
+      "grad_norm": 1.75,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.7953,
+      "mean_token_accuracy": 0.7999043371528387,
+      "num_tokens": 7138932.0,
+      "step": 250
+    },
+    {
+      "entropy": 0.7674930268898607,
+      "epoch": 0.416,
+      "grad_norm": 1.6796875,
+      "learning_rate": 5.18e-06,
+      "loss": 0.7597,
+      "mean_token_accuracy": 0.8044768057763576,
+      "num_tokens": 7420720.0,
+      "step": 260
+    },
+    {
+      "entropy": 0.7496220650151372,
+      "epoch": 0.432,
+      "grad_norm": 1.5390625,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.7468,
+      "mean_token_accuracy": 0.8073872365057468,
+      "num_tokens": 7695518.0,
+      "step": 270
+    },
+    {
+      "entropy": 0.7439631385728717,
+      "epoch": 0.448,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.7339,
+      "mean_token_accuracy": 0.8076559219509363,
+      "num_tokens": 7983543.0,
+      "step": 280
+    },
+    {
+      "entropy": 0.7457249015569687,
+      "epoch": 0.464,
+      "grad_norm": 2.65625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.741,
+      "mean_token_accuracy": 0.8087894931435585,
+      "num_tokens": 8271787.0,
+      "step": 290
+    },
+    {
+      "entropy": 0.7761816097423434,
+      "epoch": 0.48,
+      "grad_norm": 1.78125,
+      "learning_rate": 5.98e-06,
+      "loss": 0.7548,
+      "mean_token_accuracy": 0.8063571337610483,
+      "num_tokens": 8556170.0,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_math_entropy": 0.7653279647827148,
+      "eval_math_loss": 0.7636919021606445,
+      "eval_math_mean_token_accuracy": 0.8056883907318115,
+      "eval_math_num_tokens": 8556170.0,
+      "eval_math_runtime": 35.8013,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.491,
+      "step": 300
+    },
+    {
+      "epoch": 0.48,
+      "eval_physics_entropy": 0.6890090732574463,
+      "eval_physics_loss": 0.6574941277503967,
+      "eval_physics_mean_token_accuracy": 0.8227382025718689,
+      "eval_physics_num_tokens": 8556170.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 300
+    },
+    {
+      "entropy": 0.7349074123427272,
+      "epoch": 0.496,
+      "grad_norm": 1.828125,
+      "learning_rate": 6.18e-06,
+      "loss": 0.7172,
+      "mean_token_accuracy": 0.8132212206721305,
+      "num_tokens": 8840590.0,
+      "step": 310
+    },
+    {
+      "entropy": 0.7443849379196763,
+      "epoch": 0.512,
+      "grad_norm": 1.6640625,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.7181,
+      "mean_token_accuracy": 0.812743404135108,
+      "num_tokens": 9131903.0,
+      "step": 320
+    },
+    {
+      "entropy": 0.7170431386679411,
+      "epoch": 0.528,
+      "grad_norm": 1.7109375,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.7045,
+      "mean_token_accuracy": 0.8160003747791051,
+      "num_tokens": 9429519.0,
+      "step": 330
+    },
+    {
+      "entropy": 0.7328394843265414,
+      "epoch": 0.544,
+      "grad_norm": 2.234375,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.722,
+      "mean_token_accuracy": 0.8143766384571791,
+      "num_tokens": 9718247.0,
+      "step": 340
+    },
+    {
+      "entropy": 0.7620531113818287,
+      "epoch": 0.56,
+      "grad_norm": 2.453125,
+      "learning_rate": 6.98e-06,
+      "loss": 0.7395,
+      "mean_token_accuracy": 0.8070962373167276,
+      "num_tokens": 9991100.0,
+      "step": 350
+    },
+    {
+      "entropy": 0.7331368897110224,
+      "epoch": 0.576,
+      "grad_norm": 1.8515625,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.7173,
+      "mean_token_accuracy": 0.8115411166101694,
+      "num_tokens": 10276090.0,
+      "step": 360
+    },
+    {
+      "entropy": 0.7342592483386398,
+      "epoch": 0.592,
+      "grad_norm": 1.546875,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.7141,
+      "mean_token_accuracy": 0.8124774299561978,
+      "num_tokens": 10566762.0,
+      "step": 370
+    },
+    {
+      "entropy": 0.7440765531733632,
+      "epoch": 0.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 7.58e-06,
+      "loss": 0.7237,
+      "mean_token_accuracy": 0.8122259072959424,
+      "num_tokens": 10857002.0,
+      "step": 380
+    },
+    {
+      "entropy": 0.7216648861765862,
+      "epoch": 0.624,
+      "grad_norm": 2.25,
+      "learning_rate": 7.78e-06,
+      "loss": 0.7057,
+      "mean_token_accuracy": 0.8139940846711398,
+      "num_tokens": 11147326.0,
+      "step": 390
+    },
+    {
+      "entropy": 0.7119627617299557,
+      "epoch": 0.64,
+      "grad_norm": 1.5234375,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.6947,
+      "mean_token_accuracy": 0.8168974690139293,
+      "num_tokens": 11437710.0,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_math_entropy": 0.7429183626174927,
+      "eval_math_loss": 0.7330864071846008,
+      "eval_math_mean_token_accuracy": 0.8096566972732544,
+      "eval_math_num_tokens": 11437710.0,
+      "eval_math_runtime": 35.7944,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 400
+    },
+    {
+      "epoch": 0.64,
+      "eval_physics_entropy": 0.6820782516002655,
+      "eval_physics_loss": 0.6511842012405396,
+      "eval_physics_mean_token_accuracy": 0.8221303143501282,
+      "eval_physics_num_tokens": 11437710.0,
+      "eval_physics_runtime": 41.4584,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 400
+    },
+    {
+      "entropy": 0.712121813185513,
+      "epoch": 0.656,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.18e-06,
+      "loss": 0.6914,
+      "mean_token_accuracy": 0.8179911982268095,
+      "num_tokens": 11727268.0,
+      "step": 410
+    },
+    {
+      "entropy": 0.7159143410623073,
+      "epoch": 0.672,
+      "grad_norm": 1.65625,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.7016,
+      "mean_token_accuracy": 0.813546035438776,
+      "num_tokens": 12020478.0,
+      "step": 420
+    },
+    {
+      "entropy": 0.7149255678057671,
+      "epoch": 0.688,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.6908,
+      "mean_token_accuracy": 0.8162215895950794,
+      "num_tokens": 12300380.0,
+      "step": 430
+    },
+    {
+      "entropy": 0.7053035136312247,
+      "epoch": 0.704,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.78e-06,
+      "loss": 0.6839,
+      "mean_token_accuracy": 0.8172036681324244,
+      "num_tokens": 12591366.0,
+      "step": 440
+    },
+    {
+      "entropy": 0.7228720126673579,
+      "epoch": 0.72,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.98e-06,
+      "loss": 0.6945,
+      "mean_token_accuracy": 0.8145164046436548,
+      "num_tokens": 12869618.0,
+      "step": 450
+    },
+    {
+      "entropy": 0.7052439017221331,
+      "epoch": 0.736,
+      "grad_norm": 1.75,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.6866,
+      "mean_token_accuracy": 0.8176059562712907,
+      "num_tokens": 13156311.0,
+      "step": 460
+    },
+    {
+      "entropy": 0.692553037032485,
+      "epoch": 0.752,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.6644,
+      "mean_token_accuracy": 0.821098442748189,
+      "num_tokens": 13438911.0,
+      "step": 470
+    },
+    {
+      "entropy": 0.6714967370033265,
+      "epoch": 0.768,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.58e-06,
+      "loss": 0.6595,
+      "mean_token_accuracy": 0.825907539948821,
+      "num_tokens": 13735397.0,
+      "step": 480
+    },
+    {
+      "entropy": 0.7148594724014401,
+      "epoch": 0.784,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.6954,
+      "mean_token_accuracy": 0.8156244982033968,
+      "num_tokens": 14024497.0,
+      "step": 490
+    },
+    {
+      "entropy": 0.7110015576705336,
+      "epoch": 0.8,
+      "grad_norm": 1.6328125,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.69,
+      "mean_token_accuracy": 0.8156023971736431,
+      "num_tokens": 14316427.0,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_math_entropy": 0.7275117125511169,
+      "eval_math_loss": 0.7106672525405884,
+      "eval_math_mean_token_accuracy": 0.8132652158737183,
+      "eval_math_num_tokens": 14316427.0,
+      "eval_math_runtime": 35.7797,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 500
+    },
+    {
+      "epoch": 0.8,
+      "eval_physics_entropy": 0.6752480027675629,
+      "eval_physics_loss": 0.6474179029464722,
+      "eval_physics_mean_token_accuracy": 0.8222041974067688,
+      "eval_physics_num_tokens": 14316427.0,
+      "eval_physics_runtime": 41.1592,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 500
+    },
+    {
+      "entropy": 0.7031571570783853,
+      "epoch": 0.816,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.018e-05,
+      "loss": 0.6756,
+      "mean_token_accuracy": 0.8171336345374585,
+      "num_tokens": 14607136.0,
+      "step": 510
+    },
+    {
+      "entropy": 0.693414007127285,
+      "epoch": 0.832,
+      "grad_norm": 1.5859375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.6706,
+      "mean_token_accuracy": 0.8211553748697042,
+      "num_tokens": 14894084.0,
+      "step": 520
+    },
+    {
+      "entropy": 0.7212638273835182,
+      "epoch": 0.848,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.6965,
+      "mean_token_accuracy": 0.8145442601293326,
+      "num_tokens": 15177039.0,
+      "step": 530
+    },
+    {
+      "entropy": 0.7037423193454743,
+      "epoch": 0.864,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.6824,
+      "mean_token_accuracy": 0.816459146887064,
+      "num_tokens": 15463393.0,
+      "step": 540
+    },
+    {
+      "entropy": 0.7084541182965041,
+      "epoch": 0.88,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.6861,
+      "mean_token_accuracy": 0.8153413504362106,
+      "num_tokens": 15747943.0,
+      "step": 550
+    },
+    {
+      "entropy": 0.7014794830232859,
+      "epoch": 0.896,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.6844,
+      "mean_token_accuracy": 0.8173291139304638,
+      "num_tokens": 16027931.0,
+      "step": 560
+    },
+    {
+      "entropy": 0.6914853570982814,
+      "epoch": 0.912,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.6643,
+      "mean_token_accuracy": 0.8196637153625488,
+      "num_tokens": 16317127.0,
+      "step": 570
+    },
+    {
+      "entropy": 0.6866581376641989,
+      "epoch": 0.928,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.661,
+      "mean_token_accuracy": 0.8199340119957924,
+      "num_tokens": 16598757.0,
+      "step": 580
+    },
+    {
+      "entropy": 0.7026519874110818,
+      "epoch": 0.944,
+      "grad_norm": 7.0625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.6838,
+      "mean_token_accuracy": 0.8174264524132013,
+      "num_tokens": 16882236.0,
+      "step": 590
+    },
+    {
+      "entropy": 0.7053765447810292,
+      "epoch": 0.96,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.198e-05,
+      "loss": 0.6832,
+      "mean_token_accuracy": 0.8171489015221596,
+      "num_tokens": 17169495.0,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_math_entropy": 0.7025088634490967,
+      "eval_math_loss": 0.6926642656326294,
+      "eval_math_mean_token_accuracy": 0.8157670006752015,
+      "eval_math_num_tokens": 17169495.0,
+      "eval_math_runtime": 35.7558,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 600
+    },
+    {
+      "epoch": 0.96,
+      "eval_physics_entropy": 0.6707950072288513,
+      "eval_physics_loss": 0.6470632553100586,
+      "eval_physics_mean_token_accuracy": 0.8212249627113343,
+      "eval_physics_num_tokens": 17169495.0,
+      "eval_physics_runtime": 41.7459,
+      "eval_physics_samples_per_second": 11.977,
+      "eval_physics_steps_per_second": 2.994,
+      "step": 600
+    },
+    {
+      "entropy": 0.684873866662383,
+      "epoch": 0.976,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.218e-05,
+      "loss": 0.6671,
+      "mean_token_accuracy": 0.8196386702358722,
+      "num_tokens": 17454147.0,
+      "step": 610
+    },
+    {
+      "entropy": 0.7006405493244529,
+      "epoch": 0.992,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.672,
+      "mean_token_accuracy": 0.8165230687707663,
+      "num_tokens": 17742596.0,
+      "step": 620
+    },
+    {
+      "entropy": 0.6851861752569676,
+      "epoch": 1.008,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.6635,
+      "mean_token_accuracy": 0.8197994366288185,
+      "num_tokens": 18033540.0,
+      "step": 630
+    },
+    {
+      "entropy": 0.6624577900394797,
+      "epoch": 1.024,
+      "grad_norm": 1.5,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.6405,
+      "mean_token_accuracy": 0.823651035502553,
+      "num_tokens": 18316346.0,
+      "step": 640
+    },
+    {
+      "entropy": 0.6590202970430254,
+      "epoch": 1.04,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.6392,
+      "mean_token_accuracy": 0.8249083694070578,
+      "num_tokens": 18595230.0,
+      "step": 650
+    },
+    {
+      "entropy": 0.6720681777223945,
+      "epoch": 1.056,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.6481,
+      "mean_token_accuracy": 0.8219482194632292,
+      "num_tokens": 18885547.0,
+      "step": 660
+    },
+    {
+      "entropy": 0.666364430822432,
+      "epoch": 1.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.6453,
+      "mean_token_accuracy": 0.821395493671298,
+      "num_tokens": 19169083.0,
+      "step": 670
+    },
+    {
+      "entropy": 0.6612174535170198,
+      "epoch": 1.088,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.6328,
+      "mean_token_accuracy": 0.823925781622529,
+      "num_tokens": 19461465.0,
+      "step": 680
+    },
+    {
+      "entropy": 0.6591472556814552,
+      "epoch": 1.104,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.378e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8235542017966508,
+      "num_tokens": 19755027.0,
+      "step": 690
+    },
+    {
+      "entropy": 0.6446556953713298,
+      "epoch": 1.12,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.398e-05,
+      "loss": 0.6235,
+      "mean_token_accuracy": 0.8280277159065008,
+      "num_tokens": 20043293.0,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_math_entropy": 0.6619428229331971,
+      "eval_math_loss": 0.6814246773719788,
+      "eval_math_mean_token_accuracy": 0.8176272907257081,
+      "eval_math_num_tokens": 20043293.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 700
+    },
+    {
+      "epoch": 1.12,
+      "eval_physics_entropy": 0.6304002203941346,
+      "eval_physics_loss": 0.6498542428016663,
+      "eval_physics_mean_token_accuracy": 0.8209347081184387,
+      "eval_physics_num_tokens": 20043293.0,
+      "eval_physics_runtime": 41.4297,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 700
+    },
+    {
+      "entropy": 0.6416833782568574,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.6298,
+      "mean_token_accuracy": 0.8263315539807081,
+      "num_tokens": 20323010.0,
+      "step": 710
+    },
+    {
+      "entropy": 0.6587465461343527,
+      "epoch": 1.152,
+      "grad_norm": 1.578125,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.6376,
+      "mean_token_accuracy": 0.8255084995180368,
+      "num_tokens": 20613266.0,
+      "step": 720
+    },
+    {
+      "entropy": 0.6476276513189078,
+      "epoch": 1.168,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.6316,
+      "mean_token_accuracy": 0.8264493472874165,
+      "num_tokens": 20910404.0,
+      "step": 730
+    },
+    {
+      "entropy": 0.6569468161091209,
+      "epoch": 1.184,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.6336,
+      "mean_token_accuracy": 0.8247423082590103,
+      "num_tokens": 21200197.0,
+      "step": 740
+    },
+    {
+      "entropy": 0.6399336410686374,
+      "epoch": 1.2,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.498e-05,
+      "loss": 0.622,
+      "mean_token_accuracy": 0.8271243523806333,
+      "num_tokens": 21496441.0,
+      "step": 750
+    },
+    {
+      "entropy": 0.6582916142418981,
+      "epoch": 1.216,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.6353,
+      "mean_token_accuracy": 0.8234689589589834,
+      "num_tokens": 21778030.0,
+      "step": 760
+    },
+    {
+      "entropy": 0.637734330072999,
+      "epoch": 1.232,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.6159,
+      "mean_token_accuracy": 0.8288460947573185,
+      "num_tokens": 22060851.0,
+      "step": 770
+    },
+    {
+      "entropy": 0.636210915632546,
+      "epoch": 1.248,
+      "grad_norm": 1.5078125,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.6165,
+      "mean_token_accuracy": 0.82775485701859,
+      "num_tokens": 22348738.0,
+      "step": 780
+    },
+    {
+      "entropy": 0.6564677763730288,
+      "epoch": 1.264,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.6349,
+      "mean_token_accuracy": 0.8237839676439762,
+      "num_tokens": 22635039.0,
+      "step": 790
+    },
+    {
+      "entropy": 0.6571361353620887,
+      "epoch": 1.28,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.6366,
+      "mean_token_accuracy": 0.8244204126298428,
+      "num_tokens": 22915190.0,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_math_entropy": 0.6791466112136841,
+      "eval_math_loss": 0.669770359992981,
+      "eval_math_mean_token_accuracy": 0.8196616005897522,
+      "eval_math_num_tokens": 22915190.0,
+      "eval_math_runtime": 35.7574,
+      "eval_math_samples_per_second": 13.983,
+      "eval_math_steps_per_second": 3.496,
+      "step": 800
+    },
+    {
+      "epoch": 1.28,
+      "eval_physics_entropy": 0.6547831742763519,
+      "eval_physics_loss": 0.650165319442749,
+      "eval_physics_mean_token_accuracy": 0.82009166097641,
+      "eval_physics_num_tokens": 22915190.0,
+      "eval_physics_runtime": 41.4354,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 800
+    },
+    {
+      "entropy": 0.6429510954767466,
+      "epoch": 1.296,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.618e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8271825641393662,
+      "num_tokens": 23194947.0,
+      "step": 810
+    },
+    {
+      "entropy": 0.6526322312653065,
+      "epoch": 1.312,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.6414,
+      "mean_token_accuracy": 0.8243110898882151,
+      "num_tokens": 23483496.0,
+      "step": 820
+    },
+    {
+      "entropy": 0.6679502466693521,
+      "epoch": 1.328,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.658e-05,
+      "loss": 0.6441,
+      "mean_token_accuracy": 0.8223706830292941,
+      "num_tokens": 23762589.0,
+      "step": 830
+    },
+    {
+      "entropy": 0.6403571154922247,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.6222,
+      "mean_token_accuracy": 0.8268204662948847,
+      "num_tokens": 24041497.0,
+      "step": 840
+    },
+    {
+      "entropy": 0.6487369466573,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.698e-05,
+      "loss": 0.6282,
+      "mean_token_accuracy": 0.8273422639817,
+      "num_tokens": 24330591.0,
+      "step": 850
+    },
+    {
+      "entropy": 0.688577077165246,
+      "epoch": 1.376,
+      "grad_norm": 1.4453125,
+      "learning_rate": 1.718e-05,
+      "loss": 0.6651,
+      "mean_token_accuracy": 0.8164406325668097,
+      "num_tokens": 24611420.0,
+      "step": 860
+    },
+    {
+      "entropy": 0.643337251059711,
+      "epoch": 1.392,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.6209,
+      "mean_token_accuracy": 0.825714922696352,
+      "num_tokens": 24894694.0,
+      "step": 870
+    },
+    {
+      "entropy": 0.6371515756472945,
+      "epoch": 1.408,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.758e-05,
+      "loss": 0.6156,
+      "mean_token_accuracy": 0.8278428696095943,
+      "num_tokens": 25184946.0,
+      "step": 880
+    },
+    {
+      "entropy": 0.6312845427542925,
+      "epoch": 1.424,
+      "grad_norm": 1.25,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.615,
+      "mean_token_accuracy": 0.8290112767368555,
+      "num_tokens": 25474746.0,
+      "step": 890
+    },
+    {
+      "entropy": 0.6522130774334073,
+      "epoch": 1.44,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.798e-05,
+      "loss": 0.6295,
+      "mean_token_accuracy": 0.8248222030699253,
+      "num_tokens": 25758550.0,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_math_entropy": 0.6579540917873382,
+      "eval_math_loss": 0.6607276797294617,
+      "eval_math_mean_token_accuracy": 0.8207882242202759,
+      "eval_math_num_tokens": 25758550.0,
+      "eval_math_runtime": 35.7442,
+      "eval_math_samples_per_second": 13.988,
+      "eval_math_steps_per_second": 3.497,
+      "step": 900
+    },
+    {
+      "epoch": 1.44,
+      "eval_physics_entropy": 0.6424285123348236,
+      "eval_physics_loss": 0.651524543762207,
+      "eval_physics_mean_token_accuracy": 0.8197560949325562,
+      "eval_physics_num_tokens": 25758550.0,
+      "eval_physics_runtime": 41.435,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 900
+    },
+    {
+      "entropy": 0.6359886834397912,
+      "epoch": 1.456,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.6177,
+      "mean_token_accuracy": 0.8268891926854849,
+      "num_tokens": 26039862.0,
+      "step": 910
+    },
+    {
+      "entropy": 0.6429944572970271,
+      "epoch": 1.472,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8256016962230206,
+      "num_tokens": 26318666.0,
+      "step": 920
+    },
+    {
+      "entropy": 0.666933044232428,
+      "epoch": 1.488,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.858e-05,
+      "loss": 0.6455,
+      "mean_token_accuracy": 0.8214054699987173,
+      "num_tokens": 26594867.0,
+      "step": 930
+    },
+    {
+      "entropy": 0.6266062622889876,
+      "epoch": 1.504,
+      "grad_norm": 1.5,
+      "learning_rate": 1.878e-05,
+      "loss": 0.6092,
+      "mean_token_accuracy": 0.8303067740052938,
+      "num_tokens": 26887233.0,
+      "step": 940
+    },
+    {
+      "entropy": 0.659474827721715,
+      "epoch": 1.52,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.898e-05,
+      "loss": 0.6344,
+      "mean_token_accuracy": 0.8236033659428358,
+      "num_tokens": 27177448.0,
+      "step": 950
+    },
+    {
+      "entropy": 0.6371477670967579,
+      "epoch": 1.536,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.918e-05,
+      "loss": 0.6245,
+      "mean_token_accuracy": 0.8263927459716797,
+      "num_tokens": 27467857.0,
+      "step": 960
+    },
+    {
+      "entropy": 0.6363119596615434,
+      "epoch": 1.552,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.938e-05,
+      "loss": 0.6172,
+      "mean_token_accuracy": 0.8277775753289461,
+      "num_tokens": 27755138.0,
+      "step": 970
+    },
+    {
+      "entropy": 0.625205036252737,
+      "epoch": 1.568,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6047,
+      "mean_token_accuracy": 0.8324479281902313,
+      "num_tokens": 28049363.0,
+      "step": 980
+    },
+    {
+      "entropy": 0.6341174798086285,
+      "epoch": 1.584,
+      "grad_norm": 1.375,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6153,
+      "mean_token_accuracy": 0.8282931160181761,
+      "num_tokens": 28337890.0,
+      "step": 990
+    },
+    {
+      "entropy": 0.6248919799923897,
+      "epoch": 1.6,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6112,
+      "mean_token_accuracy": 0.827129240334034,
+      "num_tokens": 28626068.0,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_math_entropy": 0.6310200927257538,
+      "eval_math_loss": 0.651292085647583,
+      "eval_math_mean_token_accuracy": 0.8225455293655396,
+      "eval_math_num_tokens": 28626068.0,
+      "eval_math_runtime": 35.7544,
+      "eval_math_samples_per_second": 13.984,
+      "eval_math_steps_per_second": 3.496,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_physics_entropy": 0.6265229551792145,
+      "eval_physics_loss": 0.6552274227142334,
+      "eval_physics_mean_token_accuracy": 0.8185877275466918,
+      "eval_physics_num_tokens": 28626068.0,
+      "eval_physics_runtime": 41.1356,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 1000
+    },
+    {
+      "entropy": 0.6397603360936046,
+      "epoch": 1.616,
+      "grad_norm": 1.25,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.6266,
+      "mean_token_accuracy": 0.8261015623807907,
+      "num_tokens": 28913208.0,
+      "step": 1010
+    },
+    {
+      "entropy": 0.6517996810376644,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.995777777777778e-05,
+      "loss": 0.6284,
+      "mean_token_accuracy": 0.8228654317557812,
+      "num_tokens": 29188917.0,
+      "step": 1020
+    },
+    {
+      "entropy": 0.6239277983084321,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 1.2421875,
+      "learning_rate": 1.9935555555555557e-05,
+      "loss": 0.6068,
+      "mean_token_accuracy": 0.8318834796547889,
+      "num_tokens": 29477292.0,
+      "step": 1030
+    },
+    {
+      "entropy": 0.6378339532762766,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9913333333333335e-05,
+      "loss": 0.6129,
+      "mean_token_accuracy": 0.8274775009602309,
+      "num_tokens": 29772480.0,
+      "step": 1040
+    },
+    {
+      "entropy": 0.64110322073102,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.9891111111111112e-05,
+      "loss": 0.6244,
+      "mean_token_accuracy": 0.8272217728197575,
+      "num_tokens": 30061007.0,
+      "step": 1050
+    },
+    {
+      "entropy": 0.6449864653870463,
+      "epoch": 1.696,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.986888888888889e-05,
+      "loss": 0.6243,
+      "mean_token_accuracy": 0.8258783508092165,
+      "num_tokens": 30344415.0,
+      "step": 1060
+    },
+    {
+      "entropy": 0.6212725475430488,
+      "epoch": 1.712,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9846666666666668e-05,
+      "loss": 0.6076,
+      "mean_token_accuracy": 0.8308158092200756,
+      "num_tokens": 30640891.0,
+      "step": 1070
+    },
+    {
+      "entropy": 0.618477713689208,
+      "epoch": 1.728,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9824444444444445e-05,
+      "loss": 0.5981,
+      "mean_token_accuracy": 0.8317991364747286,
+      "num_tokens": 30921295.0,
+      "step": 1080
+    },
+    {
+      "entropy": 0.6253316847607493,
+      "epoch": 1.744,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.9802222222222226e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8302719250321389,
+      "num_tokens": 31206375.0,
+      "step": 1090
+    },
+    {
+      "entropy": 0.6203237606212497,
+      "epoch": 1.76,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.978e-05,
+      "loss": 0.6016,
+      "mean_token_accuracy": 0.829212823137641,
+      "num_tokens": 31491856.0,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_math_entropy": 0.6470205163955689,
+      "eval_math_loss": 0.6425853371620178,
+      "eval_math_mean_token_accuracy": 0.8245493578910827,
+      "eval_math_num_tokens": 31491856.0,
+      "eval_math_runtime": 35.7795,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1100
+    },
+    {
+      "epoch": 1.76,
+      "eval_physics_entropy": 0.6618543210029602,
+      "eval_physics_loss": 0.6551723480224609,
+      "eval_physics_mean_token_accuracy": 0.8183454179763794,
+      "eval_physics_num_tokens": 31491856.0,
+      "eval_physics_runtime": 41.448,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 1100
+    },
+    {
+      "entropy": 0.6285459484905005,
+      "epoch": 1.776,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.975777777777778e-05,
+      "loss": 0.608,
+      "mean_token_accuracy": 0.8288080658763647,
+      "num_tokens": 31782829.0,
+      "step": 1110
+    },
+    {
+      "entropy": 0.6362737115472555,
+      "epoch": 1.792,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.9735555555555556e-05,
+      "loss": 0.6168,
+      "mean_token_accuracy": 0.8268269985914231,
+      "num_tokens": 32070187.0,
+      "step": 1120
+    },
+    {
+      "entropy": 0.6320221189409494,
+      "epoch": 1.808,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9713333333333337e-05,
+      "loss": 0.6148,
+      "mean_token_accuracy": 0.8273419640958309,
+      "num_tokens": 32355072.0,
+      "step": 1130
+    },
+    {
+      "entropy": 0.6012085109949112,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.969111111111111e-05,
+      "loss": 0.5802,
+      "mean_token_accuracy": 0.8352209012955427,
+      "num_tokens": 32634013.0,
+      "step": 1140
+    },
+    {
+      "entropy": 0.6171128546819091,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 1.328125,
+      "learning_rate": 1.9668888888888892e-05,
+      "loss": 0.6044,
+      "mean_token_accuracy": 0.8311630170792341,
+      "num_tokens": 32917222.0,
+      "step": 1150
+    },
+    {
+      "entropy": 0.6315351065248251,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.9646666666666666e-05,
+      "loss": 0.6066,
+      "mean_token_accuracy": 0.8292891420423985,
+      "num_tokens": 33207147.0,
+      "step": 1160
+    },
+    {
+      "entropy": 0.6153346884995698,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.9624444444444447e-05,
+      "loss": 0.5954,
+      "mean_token_accuracy": 0.8320576064288616,
+      "num_tokens": 33496572.0,
+      "step": 1170
+    },
+    {
+      "entropy": 0.624875632673502,
+      "epoch": 1.888,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9602222222222225e-05,
+      "loss": 0.6059,
+      "mean_token_accuracy": 0.8299812860786915,
+      "num_tokens": 33775788.0,
+      "step": 1180
+    },
+    {
+      "entropy": 0.6256145890802145,
+      "epoch": 1.904,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.6084,
+      "mean_token_accuracy": 0.8293462209403515,
+      "num_tokens": 34052322.0,
+      "step": 1190
+    },
+    {
+      "entropy": 0.6057112148031593,
+      "epoch": 1.92,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.955777777777778e-05,
+      "loss": 0.5888,
+      "mean_token_accuracy": 0.8329556312412023,
+      "num_tokens": 34331468.0,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_math_entropy": 0.6444816522598267,
+      "eval_math_loss": 0.6337655782699585,
+      "eval_math_mean_token_accuracy": 0.8254525041580201,
+      "eval_math_num_tokens": 34331468.0,
+      "eval_math_runtime": 35.7615,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1200
+    },
+    {
+      "epoch": 1.92,
+      "eval_physics_entropy": 0.6525136830806733,
+      "eval_physics_loss": 0.6549710035324097,
+      "eval_physics_mean_token_accuracy": 0.8188423981666565,
+      "eval_physics_num_tokens": 34331468.0,
+      "eval_physics_runtime": 41.437,
+      "eval_physics_samples_per_second": 12.067,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1200
+    },
+    {
+      "entropy": 0.622246976941824,
+      "epoch": 1.936,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.9535555555555557e-05,
+      "loss": 0.6052,
+      "mean_token_accuracy": 0.8293883047997952,
+      "num_tokens": 34612743.0,
+      "step": 1210
+    },
+    {
+      "entropy": 0.6145633114501834,
+      "epoch": 1.952,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9513333333333335e-05,
+      "loss": 0.5899,
+      "mean_token_accuracy": 0.8315575309097767,
+      "num_tokens": 34891115.0,
+      "step": 1220
+    },
+    {
+      "entropy": 0.6348631424829364,
+      "epoch": 1.968,
+      "grad_norm": 3.625,
+      "learning_rate": 1.9491111111111113e-05,
+      "loss": 0.6247,
+      "mean_token_accuracy": 0.8266950808465481,
+      "num_tokens": 35179505.0,
+      "step": 1230
+    },
+    {
+      "entropy": 0.6054144280031324,
+      "epoch": 1.984,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.946888888888889e-05,
+      "loss": 0.5878,
+      "mean_token_accuracy": 0.8339846853166819,
+      "num_tokens": 35472596.0,
+      "step": 1240
+    },
+    {
+      "entropy": 0.614653673581779,
+      "epoch": 2.0,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9446666666666668e-05,
+      "loss": 0.5966,
+      "mean_token_accuracy": 0.8319407891482115,
+      "num_tokens": 35772848.0,
+      "step": 1250
+    },
+    {
+      "entropy": 0.5460906764492393,
+      "epoch": 2.016,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.9424444444444446e-05,
+      "loss": 0.5207,
+      "mean_token_accuracy": 0.844809464737773,
+      "num_tokens": 36060330.0,
+      "step": 1260
+    },
+    {
+      "entropy": 0.5267174040898681,
+      "epoch": 2.032,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9402222222222223e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8475256130099297,
+      "num_tokens": 36343894.0,
+      "step": 1270
+    },
+    {
+      "entropy": 0.5175346418283879,
+      "epoch": 2.048,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.938e-05,
+      "loss": 0.501,
+      "mean_token_accuracy": 0.8504213903099298,
+      "num_tokens": 36636410.0,
+      "step": 1280
+    },
+    {
+      "entropy": 0.5137943865731358,
+      "epoch": 2.064,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.935777777777778e-05,
+      "loss": 0.4962,
+      "mean_token_accuracy": 0.8507547531276941,
+      "num_tokens": 36925382.0,
+      "step": 1290
+    },
+    {
+      "entropy": 0.5208146266639233,
+      "epoch": 2.08,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9335555555555556e-05,
+      "loss": 0.4954,
+      "mean_token_accuracy": 0.849713422730565,
+      "num_tokens": 37208672.0,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_math_entropy": 0.5398842208385467,
+      "eval_math_loss": 0.6511635780334473,
+      "eval_math_mean_token_accuracy": 0.8244616360664367,
+      "eval_math_num_tokens": 37208672.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 1300
+    },
+    {
+      "epoch": 2.08,
+      "eval_physics_entropy": 0.5216713752746582,
+      "eval_physics_loss": 0.6896871328353882,
+      "eval_physics_mean_token_accuracy": 0.8157849154472351,
+      "eval_physics_num_tokens": 37208672.0,
+      "eval_physics_runtime": 41.4274,
+      "eval_physics_samples_per_second": 12.069,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 1300
+    },
+    {
+      "entropy": 0.5262768002226949,
+      "epoch": 2.096,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9313333333333334e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8485719960182905,
+      "num_tokens": 37491954.0,
+      "step": 1310
+    },
+    {
+      "entropy": 0.5144134562462568,
+      "epoch": 2.112,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.9291111111111115e-05,
+      "loss": 0.4932,
+      "mean_token_accuracy": 0.8491108123213053,
+      "num_tokens": 37787033.0,
+      "step": 1320
+    },
+    {
+      "entropy": 0.5125698197633028,
+      "epoch": 2.128,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.926888888888889e-05,
+      "loss": 0.4936,
+      "mean_token_accuracy": 0.8515845112502575,
+      "num_tokens": 38068955.0,
+      "step": 1330
+    },
+    {
+      "entropy": 0.5147779244929552,
+      "epoch": 2.144,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.924666666666667e-05,
+      "loss": 0.4998,
+      "mean_token_accuracy": 0.8497466515749693,
+      "num_tokens": 38359914.0,
+      "step": 1340
+    },
+    {
+      "entropy": 0.5150290129706263,
+      "epoch": 2.16,
+      "grad_norm": 1.28125,
+      "learning_rate": 1.9224444444444444e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8506554007530213,
+      "num_tokens": 38649684.0,
+      "step": 1350
+    },
+    {
+      "entropy": 0.49578714426606896,
+      "epoch": 2.176,
+      "grad_norm": 1.375,
+      "learning_rate": 1.9202222222222225e-05,
+      "loss": 0.4768,
+      "mean_token_accuracy": 0.8545470163226128,
+      "num_tokens": 38938155.0,
+      "step": 1360
+    },
+    {
+      "entropy": 0.5279102198779583,
+      "epoch": 2.192,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.918e-05,
+      "loss": 0.5109,
+      "mean_token_accuracy": 0.8461268071085215,
+      "num_tokens": 39228800.0,
+      "step": 1370
+    },
+    {
+      "entropy": 0.520401804894209,
+      "epoch": 2.208,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.915777777777778e-05,
+      "loss": 0.5095,
+      "mean_token_accuracy": 0.8481683298945427,
+      "num_tokens": 39513118.0,
+      "step": 1380
+    },
+    {
+      "entropy": 0.517220887541771,
+      "epoch": 2.224,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.9135555555555555e-05,
+      "loss": 0.4947,
+      "mean_token_accuracy": 0.8505345996469259,
+      "num_tokens": 39804212.0,
+      "step": 1390
+    },
+    {
+      "entropy": 0.5203588601201773,
+      "epoch": 2.24,
+      "grad_norm": 1.359375,
+      "learning_rate": 1.9113333333333336e-05,
+      "loss": 0.5071,
+      "mean_token_accuracy": 0.8482481308281422,
+      "num_tokens": 40092012.0,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_math_entropy": 0.5498721191883087,
+      "eval_math_loss": 0.6494505405426025,
+      "eval_math_mean_token_accuracy": 0.8248468618392945,
+      "eval_math_num_tokens": 40092012.0,
+      "eval_math_runtime": 35.7606,
+      "eval_math_samples_per_second": 13.982,
+      "eval_math_steps_per_second": 3.495,
+      "step": 1400
+    },
+    {
+      "epoch": 2.24,
+      "eval_physics_entropy": 0.5240244953632355,
+      "eval_physics_loss": 0.6975257992744446,
+      "eval_physics_mean_token_accuracy": 0.8148801970481873,
+      "eval_physics_num_tokens": 40092012.0,
+      "eval_physics_runtime": 41.7753,
+      "eval_physics_samples_per_second": 11.969,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 1400
+    },
+    {
+      "entropy": 0.5205017613247037,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.9091111111111113e-05,
+      "loss": 0.5042,
+      "mean_token_accuracy": 0.8483030397444964,
+      "num_tokens": 40380069.0,
+      "step": 1410
+    },
+    {
+      "entropy": 0.5286959325894713,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.906888888888889e-05,
+      "loss": 0.5089,
+      "mean_token_accuracy": 0.8469363551586866,
+      "num_tokens": 40655903.0,
+      "step": 1420
+    },
+    {
+      "entropy": 0.5245801825076342,
+      "epoch": 2.288,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.904666666666667e-05,
+      "loss": 0.505,
+      "mean_token_accuracy": 0.8467433623969555,
+      "num_tokens": 40944791.0,
+      "step": 1430
+    },
+    {
+      "entropy": 0.5089570162817836,
+      "epoch": 2.304,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.9024444444444446e-05,
+      "loss": 0.4927,
+      "mean_token_accuracy": 0.851346742734313,
+      "num_tokens": 41227146.0,
+      "step": 1440
+    },
+    {
+      "entropy": 0.5277796785347164,
+      "epoch": 2.32,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.9002222222222224e-05,
+      "loss": 0.5143,
+      "mean_token_accuracy": 0.8476031333208084,
+      "num_tokens": 41509322.0,
+      "step": 1450
+    },
+    {
+      "entropy": 0.5237356763333082,
+      "epoch": 2.336,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8484678287059069,
+      "num_tokens": 41792799.0,
+      "step": 1460
+    },
+    {
+      "entropy": 0.5347706735134125,
+      "epoch": 2.352,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.895777777777778e-05,
+      "loss": 0.5169,
+      "mean_token_accuracy": 0.8446083668619394,
+      "num_tokens": 42083080.0,
+      "step": 1470
+    },
+    {
+      "entropy": 0.5238756524398923,
+      "epoch": 2.368,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8935555555555556e-05,
+      "loss": 0.5074,
+      "mean_token_accuracy": 0.8478495571762323,
+      "num_tokens": 42368442.0,
+      "step": 1480
+    },
+    {
+      "entropy": 0.5398916637524962,
+      "epoch": 2.384,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.8913333333333334e-05,
+      "loss": 0.5247,
+      "mean_token_accuracy": 0.8446283016353846,
+      "num_tokens": 42663926.0,
+      "step": 1490
+    },
+    {
+      "entropy": 0.5313681181520223,
+      "epoch": 2.4,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.8891111111111115e-05,
+      "loss": 0.5162,
+      "mean_token_accuracy": 0.8469122361391783,
+      "num_tokens": 42946351.0,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_math_entropy": 0.5470261330604553,
+      "eval_math_loss": 0.6441261172294617,
+      "eval_math_mean_token_accuracy": 0.8258598875999451,
+      "eval_math_num_tokens": 42946351.0,
+      "eval_math_runtime": 35.7282,
+      "eval_math_samples_per_second": 13.995,
+      "eval_math_steps_per_second": 3.499,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4,
+      "eval_physics_entropy": 0.5238613975048065,
+      "eval_physics_loss": 0.6977487802505493,
+      "eval_physics_mean_token_accuracy": 0.81443390417099,
+      "eval_physics_num_tokens": 42946351.0,
+      "eval_physics_runtime": 41.457,
+      "eval_physics_samples_per_second": 12.061,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 1500
+    },
+    {
+      "entropy": 0.5377832099795341,
+      "epoch": 2.416,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.886888888888889e-05,
+      "loss": 0.5215,
+      "mean_token_accuracy": 0.8441088248044253,
+      "num_tokens": 43227782.0,
+      "step": 1510
+    },
+    {
+      "entropy": 0.5219327821396291,
+      "epoch": 2.432,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.884666666666667e-05,
+      "loss": 0.5065,
+      "mean_token_accuracy": 0.8481958381831646,
+      "num_tokens": 43511939.0,
+      "step": 1520
+    },
+    {
+      "entropy": 0.5146825276315212,
+      "epoch": 2.448,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.8824444444444445e-05,
+      "loss": 0.4976,
+      "mean_token_accuracy": 0.8503648530691862,
+      "num_tokens": 43811005.0,
+      "step": 1530
+    },
+    {
+      "entropy": 0.5279206298291683,
+      "epoch": 2.464,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8802222222222226e-05,
+      "loss": 0.5124,
+      "mean_token_accuracy": 0.8466742359101772,
+      "num_tokens": 44106223.0,
+      "step": 1540
+    },
+    {
+      "entropy": 0.548205547593534,
+      "epoch": 2.48,
+      "grad_norm": 1.515625,
+      "learning_rate": 1.878e-05,
+      "loss": 0.5272,
+      "mean_token_accuracy": 0.8421384803950787,
+      "num_tokens": 44379095.0,
+      "step": 1550
+    },
+    {
+      "entropy": 0.5311242727562785,
+      "epoch": 2.496,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.875777777777778e-05,
+      "loss": 0.5146,
+      "mean_token_accuracy": 0.8455508589744568,
+      "num_tokens": 44670316.0,
+      "step": 1560
+    },
+    {
+      "entropy": 0.5254273321479559,
+      "epoch": 2.512,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.873555555555556e-05,
+      "loss": 0.5098,
+      "mean_token_accuracy": 0.8476663187146187,
+      "num_tokens": 44952877.0,
+      "step": 1570
+    },
+    {
+      "entropy": 0.5215749599039554,
+      "epoch": 2.528,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8713333333333336e-05,
+      "loss": 0.5036,
+      "mean_token_accuracy": 0.8488380990922451,
+      "num_tokens": 45237653.0,
+      "step": 1580
+    },
+    {
+      "entropy": 0.5184456476941705,
+      "epoch": 2.544,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8691111111111114e-05,
+      "loss": 0.5051,
+      "mean_token_accuracy": 0.8494234204292297,
+      "num_tokens": 45524866.0,
+      "step": 1590
+    },
+    {
+      "entropy": 0.5188256312161684,
+      "epoch": 2.56,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.866888888888889e-05,
+      "loss": 0.4994,
+      "mean_token_accuracy": 0.8480058163404465,
+      "num_tokens": 45803783.0,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_math_entropy": 0.5415253615379334,
+      "eval_math_loss": 0.6402944326400757,
+      "eval_math_mean_token_accuracy": 0.8268161854743957,
+      "eval_math_num_tokens": 45803783.0,
+      "eval_math_runtime": 35.8351,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 1600
+    },
+    {
+      "epoch": 2.56,
+      "eval_physics_entropy": 0.5146132915019989,
+      "eval_physics_loss": 0.7024555802345276,
+      "eval_physics_mean_token_accuracy": 0.8143429985046386,
+      "eval_physics_num_tokens": 45803783.0,
+      "eval_physics_runtime": 41.4942,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1600
+    },
+    {
+      "entropy": 0.5206789819523692,
+      "epoch": 2.576,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.864666666666667e-05,
+      "loss": 0.5037,
+      "mean_token_accuracy": 0.8492428727447987,
+      "num_tokens": 46090713.0,
+      "step": 1610
+    },
+    {
+      "entropy": 0.5139110781252384,
+      "epoch": 2.592,
+      "grad_norm": 1.25,
+      "learning_rate": 1.8624444444444446e-05,
+      "loss": 0.4983,
+      "mean_token_accuracy": 0.8500846106559038,
+      "num_tokens": 46384084.0,
+      "step": 1620
+    },
+    {
+      "entropy": 0.5300323162227869,
+      "epoch": 2.608,
+      "grad_norm": 1.59375,
+      "learning_rate": 1.8602222222222224e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8454120934009552,
+      "num_tokens": 46667552.0,
+      "step": 1630
+    },
+    {
+      "entropy": 0.498782429471612,
+      "epoch": 2.624,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.858e-05,
+      "loss": 0.4798,
+      "mean_token_accuracy": 0.8554852791130543,
+      "num_tokens": 46961124.0,
+      "step": 1640
+    },
+    {
+      "entropy": 0.5273189852014184,
+      "epoch": 2.64,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.855777777777778e-05,
+      "loss": 0.5138,
+      "mean_token_accuracy": 0.8468208122998476,
+      "num_tokens": 47245326.0,
+      "step": 1650
+    },
+    {
+      "entropy": 0.5071797709912061,
+      "epoch": 2.656,
+      "grad_norm": 1.5,
+      "learning_rate": 1.8535555555555557e-05,
+      "loss": 0.4904,
+      "mean_token_accuracy": 0.8513832915574312,
+      "num_tokens": 47527347.0,
+      "step": 1660
+    },
+    {
+      "entropy": 0.5276256233453751,
+      "epoch": 2.672,
+      "grad_norm": 1.6484375,
+      "learning_rate": 1.8513333333333335e-05,
+      "loss": 0.5105,
+      "mean_token_accuracy": 0.8466553438454867,
+      "num_tokens": 47808167.0,
+      "step": 1670
+    },
+    {
+      "entropy": 0.5247279232367873,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.8491111111111112e-05,
+      "loss": 0.5049,
+      "mean_token_accuracy": 0.8479447506368161,
+      "num_tokens": 48092035.0,
+      "step": 1680
+    },
+    {
+      "entropy": 0.5260777238756418,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.846888888888889e-05,
+      "loss": 0.5153,
+      "mean_token_accuracy": 0.8464129194617271,
+      "num_tokens": 48386231.0,
+      "step": 1690
+    },
+    {
+      "entropy": 0.5193733531981707,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8446666666666667e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8482546083629131,
+      "num_tokens": 48663262.0,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_math_entropy": 0.555855593919754,
+      "eval_math_loss": 0.6335127949714661,
+      "eval_math_mean_token_accuracy": 0.8283462662696839,
+      "eval_math_num_tokens": 48663262.0,
+      "eval_math_runtime": 35.7836,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 1700
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "eval_physics_entropy": 0.5213737454414368,
+      "eval_physics_loss": 0.6999514698982239,
+      "eval_physics_mean_token_accuracy": 0.8152164473533631,
+      "eval_physics_num_tokens": 48663262.0,
+      "eval_physics_runtime": 41.4953,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1700
+    },
+    {
+      "entropy": 0.5167077284306287,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.842444444444445e-05,
+      "loss": 0.5014,
+      "mean_token_accuracy": 0.8498601946979761,
+      "num_tokens": 48949456.0,
+      "step": 1710
+    },
+    {
+      "entropy": 0.522574995085597,
+      "epoch": 2.752,
+      "grad_norm": 1.4921875,
+      "learning_rate": 1.8402222222222223e-05,
+      "loss": 0.5015,
+      "mean_token_accuracy": 0.8494253471493721,
+      "num_tokens": 49235751.0,
+      "step": 1720
+    },
+    {
+      "entropy": 0.5236585384234786,
+      "epoch": 2.768,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8380000000000004e-05,
+      "loss": 0.5081,
+      "mean_token_accuracy": 0.8480564054101706,
+      "num_tokens": 49524628.0,
+      "step": 1730
+    },
+    {
+      "entropy": 0.536912607960403,
+      "epoch": 2.784,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8357777777777778e-05,
+      "loss": 0.5166,
+      "mean_token_accuracy": 0.8444275613874197,
+      "num_tokens": 49799604.0,
+      "step": 1740
+    },
+    {
+      "entropy": 0.5194279940798878,
+      "epoch": 2.8,
+      "grad_norm": 1.234375,
+      "learning_rate": 1.833555555555556e-05,
+      "loss": 0.5044,
+      "mean_token_accuracy": 0.8491090513765812,
+      "num_tokens": 50079307.0,
+      "step": 1750
+    },
+    {
+      "entropy": 0.5226334543898702,
+      "epoch": 2.816,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.8313333333333333e-05,
+      "loss": 0.5013,
+      "mean_token_accuracy": 0.8487468950450421,
+      "num_tokens": 50372216.0,
+      "step": 1760
+    },
+    {
+      "entropy": 0.5100100792944431,
+      "epoch": 2.832,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.8291111111111114e-05,
+      "loss": 0.4977,
+      "mean_token_accuracy": 0.8502487774938345,
+      "num_tokens": 50657211.0,
+      "step": 1770
+    },
+    {
+      "entropy": 0.5339175367727875,
+      "epoch": 2.848,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.8268888888888888e-05,
+      "loss": 0.5176,
+      "mean_token_accuracy": 0.8462332025170326,
+      "num_tokens": 50946503.0,
+      "step": 1780
+    },
+    {
+      "entropy": 0.5188839716836811,
+      "epoch": 2.864,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.824666666666667e-05,
+      "loss": 0.5046,
+      "mean_token_accuracy": 0.8482739225029945,
+      "num_tokens": 51229298.0,
+      "step": 1790
+    },
+    {
+      "entropy": 0.52184556145221,
+      "epoch": 2.88,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.8224444444444447e-05,
+      "loss": 0.5,
+      "mean_token_accuracy": 0.8485745534300804,
+      "num_tokens": 51509622.0,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_math_entropy": 0.5322999174594879,
+      "eval_math_loss": 0.6332678198814392,
+      "eval_math_mean_token_accuracy": 0.8286352276802063,
+      "eval_math_num_tokens": 51509622.0,
+      "eval_math_runtime": 35.803,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 1800
+    },
+    {
+      "epoch": 2.88,
+      "eval_physics_entropy": 0.5120384588241577,
+      "eval_physics_loss": 0.7020947337150574,
+      "eval_physics_mean_token_accuracy": 0.8149435911178589,
+      "eval_physics_num_tokens": 51509622.0,
+      "eval_physics_runtime": 41.5069,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1800
+    },
+    {
+      "entropy": 0.5138611877337098,
+      "epoch": 2.896,
+      "grad_norm": 1.4765625,
+      "learning_rate": 1.8202222222222225e-05,
+      "loss": 0.496,
+      "mean_token_accuracy": 0.8509018927812576,
+      "num_tokens": 51790856.0,
+      "step": 1810
+    },
+    {
+      "entropy": 0.5193969309329987,
+      "epoch": 2.912,
+      "grad_norm": 1.484375,
+      "learning_rate": 1.8180000000000002e-05,
+      "loss": 0.5039,
+      "mean_token_accuracy": 0.8491050932556391,
+      "num_tokens": 52074393.0,
+      "step": 1820
+    },
+    {
+      "entropy": 0.5282592326402664,
+      "epoch": 2.928,
+      "grad_norm": 1.4140625,
+      "learning_rate": 1.815777777777778e-05,
+      "loss": 0.514,
+      "mean_token_accuracy": 0.8482961695641279,
+      "num_tokens": 52366467.0,
+      "step": 1830
+    },
+    {
+      "entropy": 0.512021827697754,
+      "epoch": 2.944,
+      "grad_norm": 1.4375,
+      "learning_rate": 1.8135555555555557e-05,
+      "loss": 0.4995,
+      "mean_token_accuracy": 0.8494243275374174,
+      "num_tokens": 52657863.0,
+      "step": 1840
+    },
+    {
+      "entropy": 0.5220472222194076,
+      "epoch": 2.96,
+      "grad_norm": 1.40625,
+      "learning_rate": 1.8113333333333335e-05,
+      "loss": 0.5045,
+      "mean_token_accuracy": 0.848219645395875,
+      "num_tokens": 52947183.0,
+      "step": 1850
+    },
+    {
+      "entropy": 0.5195147603750229,
+      "epoch": 2.976,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.8091111111111113e-05,
+      "loss": 0.5009,
+      "mean_token_accuracy": 0.8488207392394542,
+      "num_tokens": 53226984.0,
+      "step": 1860
+    },
+    {
+      "entropy": 0.5123168766498566,
+      "epoch": 2.992,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.806888888888889e-05,
+      "loss": 0.4941,
+      "mean_token_accuracy": 0.8505234856158495,
+      "num_tokens": 53515204.0,
+      "step": 1870
+    },
+    {
+      "entropy": 0.4752411070279777,
+      "epoch": 3.008,
+      "grad_norm": 1.875,
+      "learning_rate": 1.8046666666666668e-05,
+      "loss": 0.4333,
+      "mean_token_accuracy": 0.8668791756033898,
+      "num_tokens": 53800592.0,
+      "step": 1880
+    },
+    {
+      "entropy": 0.3587687503546476,
+      "epoch": 3.024,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.8024444444444445e-05,
+      "loss": 0.3423,
+      "mean_token_accuracy": 0.888729453831911,
+      "num_tokens": 54092017.0,
+      "step": 1890
+    },
+    {
+      "entropy": 0.3438520409166813,
+      "epoch": 3.04,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.8002222222222223e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8938188333064317,
+      "num_tokens": 54380158.0,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_math_entropy": 0.406534277677536,
+      "eval_math_loss": 0.7412194609642029,
+      "eval_math_mean_token_accuracy": 0.8225762667655945,
+      "eval_math_num_tokens": 54380158.0,
+      "eval_math_runtime": 35.7958,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 1900
+    },
+    {
+      "epoch": 3.04,
+      "eval_physics_entropy": 0.38361185932159425,
+      "eval_physics_loss": 0.8625921607017517,
+      "eval_physics_mean_token_accuracy": 0.8045027327537536,
+      "eval_physics_num_tokens": 54380158.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 1900
+    },
+    {
+      "entropy": 0.34686512807384134,
+      "epoch": 3.056,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.798e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8946099061518907,
+      "num_tokens": 54658093.0,
+      "step": 1910
+    },
+    {
+      "entropy": 0.34148691557347777,
+      "epoch": 3.072,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7957777777777778e-05,
+      "loss": 0.3202,
+      "mean_token_accuracy": 0.8947911463677883,
+      "num_tokens": 54949938.0,
+      "step": 1920
+    },
+    {
+      "entropy": 0.3277126123197377,
+      "epoch": 3.088,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7935555555555556e-05,
+      "loss": 0.3064,
+      "mean_token_accuracy": 0.8989322237670422,
+      "num_tokens": 55234643.0,
+      "step": 1930
+    },
+    {
+      "entropy": 0.34841074971482155,
+      "epoch": 3.104,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7913333333333337e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924810353666544,
+      "num_tokens": 55519679.0,
+      "step": 1940
+    },
+    {
+      "entropy": 0.33432568768039345,
+      "epoch": 3.12,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.789111111111111e-05,
+      "loss": 0.3163,
+      "mean_token_accuracy": 0.896980819106102,
+      "num_tokens": 55812397.0,
+      "step": 1950
+    },
+    {
+      "entropy": 0.33816060032695533,
+      "epoch": 3.136,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7868888888888892e-05,
+      "loss": 0.3182,
+      "mean_token_accuracy": 0.8948573391884566,
+      "num_tokens": 56101550.0,
+      "step": 1960
+    },
+    {
+      "entropy": 0.3384258272126317,
+      "epoch": 3.152,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7846666666666666e-05,
+      "loss": 0.3209,
+      "mean_token_accuracy": 0.8947299372404813,
+      "num_tokens": 56387662.0,
+      "step": 1970
+    },
+    {
+      "entropy": 0.3522730448283255,
+      "epoch": 3.168,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7824444444444447e-05,
+      "loss": 0.333,
+      "mean_token_accuracy": 0.892043673619628,
+      "num_tokens": 56673426.0,
+      "step": 1980
+    },
+    {
+      "entropy": 0.3521577626466751,
+      "epoch": 3.184,
+      "grad_norm": 2.25,
+      "learning_rate": 1.780222222222222e-05,
+      "loss": 0.3268,
+      "mean_token_accuracy": 0.89161705262959,
+      "num_tokens": 56957044.0,
+      "step": 1990
+    },
+    {
+      "entropy": 0.3363432651385665,
+      "epoch": 3.2,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.3174,
+      "mean_token_accuracy": 0.8954976741224527,
+      "num_tokens": 57250097.0,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_math_entropy": 0.3876710822582245,
+      "eval_math_loss": 0.7634654641151428,
+      "eval_math_mean_token_accuracy": 0.8221899409294129,
+      "eval_math_num_tokens": 57250097.0,
+      "eval_math_runtime": 35.8054,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2000
+    },
+    {
+      "epoch": 3.2,
+      "eval_physics_entropy": 0.35663985908031465,
+      "eval_physics_loss": 0.8939647674560547,
+      "eval_physics_mean_token_accuracy": 0.8043588905334472,
+      "eval_physics_num_tokens": 57250097.0,
+      "eval_physics_runtime": 41.1357,
+      "eval_physics_samples_per_second": 12.155,
+      "eval_physics_steps_per_second": 3.039,
+      "step": 2000
+    },
+    {
+      "entropy": 0.3453281221911311,
+      "epoch": 3.216,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7757777777777777e-05,
+      "loss": 0.3293,
+      "mean_token_accuracy": 0.8925925586372614,
+      "num_tokens": 57532026.0,
+      "step": 2010
+    },
+    {
+      "entropy": 0.34009323371574285,
+      "epoch": 3.232,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.7735555555555558e-05,
+      "loss": 0.3216,
+      "mean_token_accuracy": 0.895359569042921,
+      "num_tokens": 57814853.0,
+      "step": 2020
+    },
+    {
+      "entropy": 0.32144517404958606,
+      "epoch": 3.248,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7713333333333335e-05,
+      "loss": 0.3013,
+      "mean_token_accuracy": 0.8993485007435084,
+      "num_tokens": 58094190.0,
+      "step": 2030
+    },
+    {
+      "entropy": 0.3362208865582943,
+      "epoch": 3.2640000000000002,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.7691111111111113e-05,
+      "loss": 0.3179,
+      "mean_token_accuracy": 0.8959619626402855,
+      "num_tokens": 58374378.0,
+      "step": 2040
+    },
+    {
+      "entropy": 0.34206875674426557,
+      "epoch": 3.2800000000000002,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.766888888888889e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8937642563134431,
+      "num_tokens": 58652789.0,
+      "step": 2050
+    },
+    {
+      "entropy": 0.3506787528283894,
+      "epoch": 3.296,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7646666666666668e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8917676325887441,
+      "num_tokens": 58930014.0,
+      "step": 2060
+    },
+    {
+      "entropy": 0.345881634298712,
+      "epoch": 3.312,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7624444444444446e-05,
+      "loss": 0.3262,
+      "mean_token_accuracy": 0.8924973297864198,
+      "num_tokens": 59210270.0,
+      "step": 2070
+    },
+    {
+      "entropy": 0.34120792699977753,
+      "epoch": 3.328,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7602222222222223e-05,
+      "loss": 0.3192,
+      "mean_token_accuracy": 0.8952000044286251,
+      "num_tokens": 59496408.0,
+      "step": 2080
+    },
+    {
+      "entropy": 0.3278652913868427,
+      "epoch": 3.344,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.758e-05,
+      "loss": 0.3112,
+      "mean_token_accuracy": 0.8966204669326544,
+      "num_tokens": 59779722.0,
+      "step": 2090
+    },
+    {
+      "entropy": 0.3657307820394635,
+      "epoch": 3.36,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.755777777777778e-05,
+      "loss": 0.3419,
+      "mean_token_accuracy": 0.8882275156676769,
+      "num_tokens": 60061562.0,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_math_entropy": 0.3830914467573166,
+      "eval_math_loss": 0.7686993479728699,
+      "eval_math_mean_token_accuracy": 0.8224415044784545,
+      "eval_math_num_tokens": 60061562.0,
+      "eval_math_runtime": 35.8062,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2100
+    },
+    {
+      "epoch": 3.36,
+      "eval_physics_entropy": 0.3522386626005173,
+      "eval_physics_loss": 0.9039635062217712,
+      "eval_physics_mean_token_accuracy": 0.8043717379570008,
+      "eval_physics_num_tokens": 60061562.0,
+      "eval_physics_runtime": 41.4883,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2100
+    },
+    {
+      "entropy": 0.35135511131957176,
+      "epoch": 3.376,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7535555555555556e-05,
+      "loss": 0.3328,
+      "mean_token_accuracy": 0.8913474876433611,
+      "num_tokens": 60346574.0,
+      "step": 2110
+    },
+    {
+      "entropy": 0.3290985696949065,
+      "epoch": 3.392,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.7513333333333334e-05,
+      "loss": 0.3122,
+      "mean_token_accuracy": 0.8976335555315018,
+      "num_tokens": 60636820.0,
+      "step": 2120
+    },
+    {
+      "entropy": 0.3349749938584864,
+      "epoch": 3.408,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.749111111111111e-05,
+      "loss": 0.3148,
+      "mean_token_accuracy": 0.8964381668716669,
+      "num_tokens": 60922033.0,
+      "step": 2130
+    },
+    {
+      "entropy": 0.3461784217506647,
+      "epoch": 3.424,
+      "grad_norm": 2.25,
+      "learning_rate": 1.746888888888889e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8919767573475837,
+      "num_tokens": 61209156.0,
+      "step": 2140
+    },
+    {
+      "entropy": 0.3424711484462023,
+      "epoch": 3.44,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.7446666666666667e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8935233734548091,
+      "num_tokens": 61491228.0,
+      "step": 2150
+    },
+    {
+      "entropy": 0.3479755653068423,
+      "epoch": 3.456,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7424444444444444e-05,
+      "loss": 0.3292,
+      "mean_token_accuracy": 0.8924077559262514,
+      "num_tokens": 61773033.0,
+      "step": 2160
+    },
+    {
+      "entropy": 0.3536012376658618,
+      "epoch": 3.472,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7402222222222222e-05,
+      "loss": 0.3402,
+      "mean_token_accuracy": 0.8894314222037792,
+      "num_tokens": 62061802.0,
+      "step": 2170
+    },
+    {
+      "entropy": 0.34079490676522256,
+      "epoch": 3.488,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.3233,
+      "mean_token_accuracy": 0.8940149489790201,
+      "num_tokens": 62347490.0,
+      "step": 2180
+    },
+    {
+      "entropy": 0.34653188129886986,
+      "epoch": 3.504,
+      "grad_norm": 2.0,
+      "learning_rate": 1.735777777777778e-05,
+      "loss": 0.3243,
+      "mean_token_accuracy": 0.8935138266533613,
+      "num_tokens": 62636811.0,
+      "step": 2190
+    },
+    {
+      "entropy": 0.3420496225357056,
+      "epoch": 3.52,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.7335555555555558e-05,
+      "loss": 0.3255,
+      "mean_token_accuracy": 0.8933457151055336,
+      "num_tokens": 62928536.0,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_math_entropy": 0.39784613919258116,
+      "eval_math_loss": 0.7561159133911133,
+      "eval_math_mean_token_accuracy": 0.8226356348991394,
+      "eval_math_num_tokens": 62928536.0,
+      "eval_math_runtime": 35.8487,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 2200
+    },
+    {
+      "epoch": 3.52,
+      "eval_physics_entropy": 0.3542559086084366,
+      "eval_physics_loss": 0.8993895053863525,
+      "eval_physics_mean_token_accuracy": 0.8046151103973389,
+      "eval_physics_num_tokens": 62928536.0,
+      "eval_physics_runtime": 41.5136,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2200
+    },
+    {
+      "entropy": 0.3441290996037424,
+      "epoch": 3.536,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.7313333333333336e-05,
+      "loss": 0.3257,
+      "mean_token_accuracy": 0.8930490992963314,
+      "num_tokens": 63212496.0,
+      "step": 2210
+    },
+    {
+      "entropy": 0.3405540899373591,
+      "epoch": 3.552,
+      "grad_norm": 2.0,
+      "learning_rate": 1.7291111111111113e-05,
+      "loss": 0.3211,
+      "mean_token_accuracy": 0.8948382496833801,
+      "num_tokens": 63506921.0,
+      "step": 2220
+    },
+    {
+      "entropy": 0.33624334512278437,
+      "epoch": 3.568,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.726888888888889e-05,
+      "loss": 0.3159,
+      "mean_token_accuracy": 0.8952401872724295,
+      "num_tokens": 63803208.0,
+      "step": 2230
+    },
+    {
+      "entropy": 0.34232397861778735,
+      "epoch": 3.584,
+      "grad_norm": 2.25,
+      "learning_rate": 1.724666666666667e-05,
+      "loss": 0.3238,
+      "mean_token_accuracy": 0.8939516879618168,
+      "num_tokens": 64084867.0,
+      "step": 2240
+    },
+    {
+      "entropy": 0.33109099259600044,
+      "epoch": 3.6,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7224444444444446e-05,
+      "loss": 0.3153,
+      "mean_token_accuracy": 0.8967003002762794,
+      "num_tokens": 64377046.0,
+      "step": 2250
+    },
+    {
+      "entropy": 0.35013388488441705,
+      "epoch": 3.616,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7202222222222224e-05,
+      "loss": 0.331,
+      "mean_token_accuracy": 0.8905543919652701,
+      "num_tokens": 64664301.0,
+      "step": 2260
+    },
+    {
+      "entropy": 0.3353199644945562,
+      "epoch": 3.632,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.718e-05,
+      "loss": 0.316,
+      "mean_token_accuracy": 0.895779125764966,
+      "num_tokens": 64946172.0,
+      "step": 2270
+    },
+    {
+      "entropy": 0.3319018458016217,
+      "epoch": 3.648,
+      "grad_norm": 2.0,
+      "learning_rate": 1.715777777777778e-05,
+      "loss": 0.3138,
+      "mean_token_accuracy": 0.8972285516560078,
+      "num_tokens": 65230709.0,
+      "step": 2280
+    },
+    {
+      "entropy": 0.36236064694821835,
+      "epoch": 3.664,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7135555555555557e-05,
+      "loss": 0.344,
+      "mean_token_accuracy": 0.8877242401242256,
+      "num_tokens": 65515738.0,
+      "step": 2290
+    },
+    {
+      "entropy": 0.3466769499704242,
+      "epoch": 3.68,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7113333333333334e-05,
+      "loss": 0.3276,
+      "mean_token_accuracy": 0.8919828839600086,
+      "num_tokens": 65802680.0,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_math_entropy": 0.3900481255054474,
+      "eval_math_loss": 0.7627390623092651,
+      "eval_math_mean_token_accuracy": 0.8218873319625855,
+      "eval_math_num_tokens": 65802680.0,
+      "eval_math_runtime": 35.8389,
+      "eval_math_samples_per_second": 13.951,
+      "eval_math_steps_per_second": 3.488,
+      "step": 2300
+    },
+    {
+      "epoch": 3.68,
+      "eval_physics_entropy": 0.3501615208387375,
+      "eval_physics_loss": 0.9098384380340576,
+      "eval_physics_mean_token_accuracy": 0.8042201838493347,
+      "eval_physics_num_tokens": 65802680.0,
+      "eval_physics_runtime": 41.525,
+      "eval_physics_samples_per_second": 12.041,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 2300
+    },
+    {
+      "entropy": 0.3516978560946882,
+      "epoch": 3.6959999999999997,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.7091111111111112e-05,
+      "loss": 0.3334,
+      "mean_token_accuracy": 0.8908947184681892,
+      "num_tokens": 66089104.0,
+      "step": 2310
+    },
+    {
+      "entropy": 0.3427901498042047,
+      "epoch": 3.7119999999999997,
+      "grad_norm": 2.0,
+      "learning_rate": 1.706888888888889e-05,
+      "loss": 0.3228,
+      "mean_token_accuracy": 0.8933374501764775,
+      "num_tokens": 66369895.0,
+      "step": 2320
+    },
+    {
+      "entropy": 0.3467458762228489,
+      "epoch": 3.7279999999999998,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.704666666666667e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8924272887408733,
+      "num_tokens": 66653165.0,
+      "step": 2330
+    },
+    {
+      "entropy": 0.34478482212871314,
+      "epoch": 3.7439999999999998,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.7024444444444445e-05,
+      "loss": 0.3245,
+      "mean_token_accuracy": 0.8931741625070572,
+      "num_tokens": 66941476.0,
+      "step": 2340
+    },
+    {
+      "entropy": 0.34707736801356076,
+      "epoch": 3.76,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.7002222222222226e-05,
+      "loss": 0.3306,
+      "mean_token_accuracy": 0.8919444873929023,
+      "num_tokens": 67225380.0,
+      "step": 2350
+    },
+    {
+      "entropy": 0.35979456212371586,
+      "epoch": 3.776,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.698e-05,
+      "loss": 0.3425,
+      "mean_token_accuracy": 0.888241409137845,
+      "num_tokens": 67516641.0,
+      "step": 2360
+    },
+    {
+      "entropy": 0.337408459559083,
+      "epoch": 3.792,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.695777777777778e-05,
+      "loss": 0.3207,
+      "mean_token_accuracy": 0.8952056996524334,
+      "num_tokens": 67808009.0,
+      "step": 2370
+    },
+    {
+      "entropy": 0.35541942873969673,
+      "epoch": 3.808,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6935555555555555e-05,
+      "loss": 0.3354,
+      "mean_token_accuracy": 0.8904807876795531,
+      "num_tokens": 68092996.0,
+      "step": 2380
+    },
+    {
+      "entropy": 0.338986696396023,
+      "epoch": 3.824,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.6913333333333336e-05,
+      "loss": 0.3217,
+      "mean_token_accuracy": 0.8948666993528605,
+      "num_tokens": 68380631.0,
+      "step": 2390
+    },
+    {
+      "entropy": 0.34510293062776326,
+      "epoch": 3.84,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.689111111111111e-05,
+      "loss": 0.327,
+      "mean_token_accuracy": 0.8922771938145161,
+      "num_tokens": 68666505.0,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_math_entropy": 0.3931693367958069,
+      "eval_math_loss": 0.7599309682846069,
+      "eval_math_mean_token_accuracy": 0.8226110072135925,
+      "eval_math_num_tokens": 68666505.0,
+      "eval_math_runtime": 35.819,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 2400
+    },
+    {
+      "epoch": 3.84,
+      "eval_physics_entropy": 0.3486952732801437,
+      "eval_physics_loss": 0.9106001257896423,
+      "eval_physics_mean_token_accuracy": 0.8040544500350952,
+      "eval_physics_num_tokens": 68666505.0,
+      "eval_physics_runtime": 41.4809,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2400
+    },
+    {
+      "entropy": 0.341458183247596,
+      "epoch": 3.856,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.686888888888889e-05,
+      "loss": 0.3259,
+      "mean_token_accuracy": 0.8934879627078771,
+      "num_tokens": 68951258.0,
+      "step": 2410
+    },
+    {
+      "entropy": 0.35674845492467283,
+      "epoch": 3.872,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.684666666666667e-05,
+      "loss": 0.3383,
+      "mean_token_accuracy": 0.8894801624119282,
+      "num_tokens": 69243705.0,
+      "step": 2420
+    },
+    {
+      "entropy": 0.3472815524786711,
+      "epoch": 3.888,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.6824444444444447e-05,
+      "loss": 0.3299,
+      "mean_token_accuracy": 0.8912571128457785,
+      "num_tokens": 69529393.0,
+      "step": 2430
+    },
+    {
+      "entropy": 0.345150712877512,
+      "epoch": 3.904,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6802222222222224e-05,
+      "loss": 0.3269,
+      "mean_token_accuracy": 0.8925980541855096,
+      "num_tokens": 69820950.0,
+      "step": 2440
+    },
+    {
+      "entropy": 0.3369919521734118,
+      "epoch": 3.92,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.317,
+      "mean_token_accuracy": 0.8948526129126548,
+      "num_tokens": 70097736.0,
+      "step": 2450
+    },
+    {
+      "entropy": 0.3590950695797801,
+      "epoch": 3.936,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.675777777777778e-05,
+      "loss": 0.34,
+      "mean_token_accuracy": 0.8890990480780602,
+      "num_tokens": 70395123.0,
+      "step": 2460
+    },
+    {
+      "entropy": 0.345857547596097,
+      "epoch": 3.952,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.6735555555555557e-05,
+      "loss": 0.3301,
+      "mean_token_accuracy": 0.8929780598729848,
+      "num_tokens": 70678153.0,
+      "step": 2470
+    },
+    {
+      "entropy": 0.3558691343292594,
+      "epoch": 3.968,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6713333333333335e-05,
+      "loss": 0.3375,
+      "mean_token_accuracy": 0.8891723170876503,
+      "num_tokens": 70963030.0,
+      "step": 2480
+    },
+    {
+      "entropy": 0.33871191181242466,
+      "epoch": 3.984,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.6691111111111112e-05,
+      "loss": 0.3215,
+      "mean_token_accuracy": 0.8948656614869833,
+      "num_tokens": 71264525.0,
+      "step": 2490
+    },
+    {
+      "entropy": 0.3556901432573795,
+      "epoch": 4.0,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.666888888888889e-05,
+      "loss": 0.337,
+      "mean_token_accuracy": 0.8900394972413779,
+      "num_tokens": 71545696.0,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_math_entropy": 0.3949365179538727,
+      "eval_math_loss": 0.7542520761489868,
+      "eval_math_mean_token_accuracy": 0.8234356184005738,
+      "eval_math_num_tokens": 71545696.0,
+      "eval_math_runtime": 35.7966,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 2500
+    },
+    {
+      "epoch": 4.0,
+      "eval_physics_entropy": 0.3519064650535583,
+      "eval_physics_loss": 0.9068982005119324,
+      "eval_physics_mean_token_accuracy": 0.8042608909606933,
+      "eval_physics_num_tokens": 71545696.0,
+      "eval_physics_runtime": 41.1579,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 2500
+    },
+    {
+      "entropy": 0.22543487511575222,
+      "epoch": 4.016,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.6646666666666668e-05,
+      "loss": 0.1799,
+      "mean_token_accuracy": 0.9439745012670755,
+      "num_tokens": 71839909.0,
+      "step": 2510
+    },
+    {
+      "entropy": 0.2174504542723298,
+      "epoch": 4.032,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.6624444444444445e-05,
+      "loss": 0.1708,
+      "mean_token_accuracy": 0.9454751797020435,
+      "num_tokens": 72128860.0,
+      "step": 2520
+    },
+    {
+      "entropy": 0.20308812106959523,
+      "epoch": 4.048,
+      "grad_norm": 1.75,
+      "learning_rate": 1.6602222222222223e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9484880570322275,
+      "num_tokens": 72415886.0,
+      "step": 2530
+    },
+    {
+      "entropy": 0.19942862098105252,
+      "epoch": 4.064,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.658e-05,
+      "loss": 0.1599,
+      "mean_token_accuracy": 0.9482992067933083,
+      "num_tokens": 72695806.0,
+      "step": 2540
+    },
+    {
+      "entropy": 0.19497004388831557,
+      "epoch": 4.08,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.6557777777777778e-05,
+      "loss": 0.1524,
+      "mean_token_accuracy": 0.9501827869564294,
+      "num_tokens": 72986978.0,
+      "step": 2550
+    },
+    {
+      "entropy": 0.1978817671071738,
+      "epoch": 4.096,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.6535555555555556e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9481502700597048,
+      "num_tokens": 73271299.0,
+      "step": 2560
+    },
+    {
+      "entropy": 0.19556209091097115,
+      "epoch": 4.112,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6513333333333333e-05,
+      "loss": 0.1569,
+      "mean_token_accuracy": 0.9491645645350217,
+      "num_tokens": 73562479.0,
+      "step": 2570
+    },
+    {
+      "entropy": 0.19597707632929087,
+      "epoch": 4.128,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.6491111111111114e-05,
+      "loss": 0.1588,
+      "mean_token_accuracy": 0.9492202006280422,
+      "num_tokens": 73845445.0,
+      "step": 2580
+    },
+    {
+      "entropy": 0.19680100358091296,
+      "epoch": 4.144,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.646888888888889e-05,
+      "loss": 0.1611,
+      "mean_token_accuracy": 0.9478188194334507,
+      "num_tokens": 74135425.0,
+      "step": 2590
+    },
+    {
+      "entropy": 0.20560383140109478,
+      "epoch": 4.16,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.644666666666667e-05,
+      "loss": 0.1678,
+      "mean_token_accuracy": 0.945937181636691,
+      "num_tokens": 74431943.0,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_math_entropy": 0.2884341893196106,
+      "eval_math_loss": 0.9303049445152283,
+      "eval_math_mean_token_accuracy": 0.8163944878578187,
+      "eval_math_num_tokens": 74431943.0,
+      "eval_math_runtime": 35.7808,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 2600
+    },
+    {
+      "epoch": 4.16,
+      "eval_physics_entropy": 0.2818371082544327,
+      "eval_physics_loss": 1.0674083232879639,
+      "eval_physics_mean_token_accuracy": 0.7992216634750366,
+      "eval_physics_num_tokens": 74431943.0,
+      "eval_physics_runtime": 41.4876,
+      "eval_physics_samples_per_second": 12.052,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 2600
+    },
+    {
+      "entropy": 0.20040173502638936,
+      "epoch": 4.176,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6424444444444444e-05,
+      "loss": 0.1641,
+      "mean_token_accuracy": 0.947614687308669,
+      "num_tokens": 74713404.0,
+      "step": 2610
+    },
+    {
+      "entropy": 0.19675334333442152,
+      "epoch": 4.192,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.6402222222222225e-05,
+      "loss": 0.1587,
+      "mean_token_accuracy": 0.948599212616682,
+      "num_tokens": 74997493.0,
+      "step": 2620
+    },
+    {
+      "entropy": 0.19941296209581197,
+      "epoch": 4.208,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.638e-05,
+      "loss": 0.1591,
+      "mean_token_accuracy": 0.94874595887959,
+      "num_tokens": 75275375.0,
+      "step": 2630
+    },
+    {
+      "entropy": 0.19698526146821677,
+      "epoch": 4.224,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.635777777777778e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9489138159900904,
+      "num_tokens": 75563461.0,
+      "step": 2640
+    },
+    {
+      "entropy": 0.2004603309556842,
+      "epoch": 4.24,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6335555555555558e-05,
+      "loss": 0.1662,
+      "mean_token_accuracy": 0.9469812735915184,
+      "num_tokens": 75854808.0,
+      "step": 2650
+    },
+    {
+      "entropy": 0.19907335569150747,
+      "epoch": 4.256,
+      "grad_norm": 2.21875,
+      "learning_rate": 1.6313333333333335e-05,
+      "loss": 0.1629,
+      "mean_token_accuracy": 0.9474829658865929,
+      "num_tokens": 76148449.0,
+      "step": 2660
+    },
+    {
+      "entropy": 0.19561562612652778,
+      "epoch": 4.272,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.6291111111111113e-05,
+      "loss": 0.1579,
+      "mean_token_accuracy": 0.9487213175743818,
+      "num_tokens": 76435793.0,
+      "step": 2670
+    },
+    {
+      "entropy": 0.19496299759484828,
+      "epoch": 4.288,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.626888888888889e-05,
+      "loss": 0.1597,
+      "mean_token_accuracy": 0.9485157284885645,
+      "num_tokens": 76726162.0,
+      "step": 2680
+    },
+    {
+      "entropy": 0.20333012314513327,
+      "epoch": 4.304,
+      "grad_norm": 2.125,
+      "learning_rate": 1.6246666666666668e-05,
+      "loss": 0.1668,
+      "mean_token_accuracy": 0.94562135823071,
+      "num_tokens": 77016858.0,
+      "step": 2690
+    },
+    {
+      "entropy": 0.19900708929635585,
+      "epoch": 4.32,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.6224444444444446e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9465276099741459,
+      "num_tokens": 77311424.0,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_math_entropy": 0.2942132350206375,
+      "eval_math_loss": 0.9334467053413391,
+      "eval_math_mean_token_accuracy": 0.8159489192962647,
+      "eval_math_num_tokens": 77311424.0,
+      "eval_math_runtime": 35.8238,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 2700
+    },
+    {
+      "epoch": 4.32,
+      "eval_physics_entropy": 0.28400694143772126,
+      "eval_physics_loss": 1.0711963176727295,
+      "eval_physics_mean_token_accuracy": 0.7983153300285339,
+      "eval_physics_num_tokens": 77311424.0,
+      "eval_physics_runtime": 41.5147,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 2700
+    },
+    {
+      "entropy": 0.20124221742153167,
+      "epoch": 4.336,
+      "grad_norm": 2.1875,
+      "learning_rate": 1.6202222222222223e-05,
+      "loss": 0.1659,
+      "mean_token_accuracy": 0.9459259379655123,
+      "num_tokens": 77596829.0,
+      "step": 2710
+    },
+    {
+      "entropy": 0.20279881260357796,
+      "epoch": 4.352,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.618e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9458406940102577,
+      "num_tokens": 77889107.0,
+      "step": 2720
+    },
+    {
+      "entropy": 0.20401687268167734,
+      "epoch": 4.368,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.615777777777778e-05,
+      "loss": 0.1683,
+      "mean_token_accuracy": 0.9460482109338045,
+      "num_tokens": 78176273.0,
+      "step": 2730
+    },
+    {
+      "entropy": 0.20603159624151884,
+      "epoch": 4.384,
+      "grad_norm": 2.25,
+      "learning_rate": 1.6135555555555556e-05,
+      "loss": 0.171,
+      "mean_token_accuracy": 0.9445526108145714,
+      "num_tokens": 78461335.0,
+      "step": 2740
+    },
+    {
+      "entropy": 0.20936214146204293,
+      "epoch": 4.4,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6113333333333334e-05,
+      "loss": 0.1746,
+      "mean_token_accuracy": 0.943215336278081,
+      "num_tokens": 78748521.0,
+      "step": 2750
+    },
+    {
+      "entropy": 0.20285641457885503,
+      "epoch": 4.416,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.609111111111111e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.9455374360084534,
+      "num_tokens": 79028488.0,
+      "step": 2760
+    },
+    {
+      "entropy": 0.1989856549538672,
+      "epoch": 4.432,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.606888888888889e-05,
+      "loss": 0.162,
+      "mean_token_accuracy": 0.9475978959351778,
+      "num_tokens": 79315788.0,
+      "step": 2770
+    },
+    {
+      "entropy": 0.20125137683935462,
+      "epoch": 4.448,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.6046666666666667e-05,
+      "loss": 0.1667,
+      "mean_token_accuracy": 0.9461831945925951,
+      "num_tokens": 79595473.0,
+      "step": 2780
+    },
+    {
+      "entropy": 0.20731261624023317,
+      "epoch": 4.464,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.6024444444444444e-05,
+      "loss": 0.1719,
+      "mean_token_accuracy": 0.9443279825150966,
+      "num_tokens": 79884376.0,
+      "step": 2790
+    },
+    {
+      "entropy": 0.20476281307637692,
+      "epoch": 4.48,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.6002222222222222e-05,
+      "loss": 0.1676,
+      "mean_token_accuracy": 0.9456661235541105,
+      "num_tokens": 80163433.0,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_math_entropy": 0.2912301456928253,
+      "eval_math_loss": 0.9387769103050232,
+      "eval_math_mean_token_accuracy": 0.8159988284111023,
+      "eval_math_num_tokens": 80163433.0,
+      "eval_math_runtime": 35.8056,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 2800
+    },
+    {
+      "epoch": 4.48,
+      "eval_physics_entropy": 0.283928528547287,
+      "eval_physics_loss": 1.0744123458862305,
+      "eval_physics_mean_token_accuracy": 0.79825497341156,
+      "eval_physics_num_tokens": 80163433.0,
+      "eval_physics_runtime": 41.5025,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2800
+    },
+    {
+      "entropy": 0.21272651255130767,
+      "epoch": 4.496,
+      "grad_norm": 2.484375,
+      "learning_rate": 1.5980000000000003e-05,
+      "loss": 0.1786,
+      "mean_token_accuracy": 0.94269258081913,
+      "num_tokens": 80443676.0,
+      "step": 2810
+    },
+    {
+      "entropy": 0.20025675450451672,
+      "epoch": 4.5120000000000005,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.5957777777777777e-05,
+      "loss": 0.165,
+      "mean_token_accuracy": 0.9459723506122828,
+      "num_tokens": 80731924.0,
+      "step": 2820
+    },
+    {
+      "entropy": 0.20602681376039983,
+      "epoch": 4.5280000000000005,
+      "grad_norm": 2.203125,
+      "learning_rate": 1.5935555555555558e-05,
+      "loss": 0.1714,
+      "mean_token_accuracy": 0.9448773771524429,
+      "num_tokens": 81022957.0,
+      "step": 2830
+    },
+    {
+      "entropy": 0.19887811122462154,
+      "epoch": 4.5440000000000005,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.5913333333333332e-05,
+      "loss": 0.1649,
+      "mean_token_accuracy": 0.9466291036456823,
+      "num_tokens": 81309913.0,
+      "step": 2840
+    },
+    {
+      "entropy": 0.20165206529200078,
+      "epoch": 4.5600000000000005,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5891111111111113e-05,
+      "loss": 0.1685,
+      "mean_token_accuracy": 0.9454752709716558,
+      "num_tokens": 81589202.0,
+      "step": 2850
+    },
+    {
+      "entropy": 0.2059997975360602,
+      "epoch": 4.576,
+      "grad_norm": 2.375,
+      "learning_rate": 1.5868888888888888e-05,
+      "loss": 0.1704,
+      "mean_token_accuracy": 0.9445979256182909,
+      "num_tokens": 81869804.0,
+      "step": 2860
+    },
+    {
+      "entropy": 0.2071733211632818,
+      "epoch": 4.592,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.584666666666667e-05,
+      "loss": 0.1733,
+      "mean_token_accuracy": 0.9441272236406804,
+      "num_tokens": 82148464.0,
+      "step": 2870
+    },
+    {
+      "entropy": 0.20506689832545816,
+      "epoch": 4.608,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5824444444444446e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9445500578731298,
+      "num_tokens": 82427716.0,
+      "step": 2880
+    },
+    {
+      "entropy": 0.20488002435304226,
+      "epoch": 4.624,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5802222222222224e-05,
+      "loss": 0.1696,
+      "mean_token_accuracy": 0.9450069412589073,
+      "num_tokens": 82710604.0,
+      "step": 2890
+    },
+    {
+      "entropy": 0.20467088604345918,
+      "epoch": 4.64,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.578e-05,
+      "loss": 0.1688,
+      "mean_token_accuracy": 0.944780471175909,
+      "num_tokens": 82990401.0,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_math_entropy": 0.29218543314933776,
+      "eval_math_loss": 0.9344849586486816,
+      "eval_math_mean_token_accuracy": 0.816554964542389,
+      "eval_math_num_tokens": 82990401.0,
+      "eval_math_runtime": 35.7784,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 2900
+    },
+    {
+      "epoch": 4.64,
+      "eval_physics_entropy": 0.2805333559513092,
+      "eval_physics_loss": 1.0783778429031372,
+      "eval_physics_mean_token_accuracy": 0.7986170620918274,
+      "eval_physics_num_tokens": 82990401.0,
+      "eval_physics_runtime": 41.499,
+      "eval_physics_samples_per_second": 12.048,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 2900
+    },
+    {
+      "entropy": 0.20116342874243856,
+      "epoch": 4.656,
+      "grad_norm": 2.421875,
+      "learning_rate": 1.575777777777778e-05,
+      "loss": 0.1639,
+      "mean_token_accuracy": 0.9467363331466914,
+      "num_tokens": 83269847.0,
+      "step": 2910
+    },
+    {
+      "entropy": 0.20632342710159718,
+      "epoch": 4.672,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5735555555555557e-05,
+      "loss": 0.1709,
+      "mean_token_accuracy": 0.9442934289574623,
+      "num_tokens": 83550833.0,
+      "step": 2920
+    },
+    {
+      "entropy": 0.2032525334507227,
+      "epoch": 4.688,
+      "grad_norm": 2.15625,
+      "learning_rate": 1.5713333333333334e-05,
+      "loss": 0.1693,
+      "mean_token_accuracy": 0.9453159097582102,
+      "num_tokens": 83840860.0,
+      "step": 2930
+    },
+    {
+      "entropy": 0.20695003918372096,
+      "epoch": 4.704,
+      "grad_norm": 2.234375,
+      "learning_rate": 1.5691111111111112e-05,
+      "loss": 0.1735,
+      "mean_token_accuracy": 0.9437373761087656,
+      "num_tokens": 84128625.0,
+      "step": 2940
+    },
+    {
+      "entropy": 0.208350771619007,
+      "epoch": 4.72,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.5668888888888893e-05,
+      "loss": 0.1736,
+      "mean_token_accuracy": 0.9436807181686163,
+      "num_tokens": 84423857.0,
+      "step": 2950
+    },
+    {
+      "entropy": 0.20460730344057082,
+      "epoch": 4.736,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5646666666666667e-05,
+      "loss": 0.1697,
+      "mean_token_accuracy": 0.9452991545200348,
+      "num_tokens": 84695402.0,
+      "step": 2960
+    },
+    {
+      "entropy": 0.19873600411228837,
+      "epoch": 4.752,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.5624444444444448e-05,
+      "loss": 0.1635,
+      "mean_token_accuracy": 0.9467461623251439,
+      "num_tokens": 84980763.0,
+      "step": 2970
+    },
+    {
+      "entropy": 0.203874301770702,
+      "epoch": 4.768,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5602222222222222e-05,
+      "loss": 0.1679,
+      "mean_token_accuracy": 0.945096442475915,
+      "num_tokens": 85272835.0,
+      "step": 2980
+    },
+    {
+      "entropy": 0.2079331438988447,
+      "epoch": 4.784,
+      "grad_norm": 2.359375,
+      "learning_rate": 1.5580000000000003e-05,
+      "loss": 0.1767,
+      "mean_token_accuracy": 0.9423614408820867,
+      "num_tokens": 85568674.0,
+      "step": 2990
+    },
+    {
+      "entropy": 0.20461545852012933,
+      "epoch": 4.8,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.5557777777777778e-05,
+      "loss": 0.1701,
+      "mean_token_accuracy": 0.9448350362479687,
+      "num_tokens": 85856157.0,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_math_entropy": 0.29029106283187867,
+      "eval_math_loss": 0.9365277290344238,
+      "eval_math_mean_token_accuracy": 0.8171119575500488,
+      "eval_math_num_tokens": 85856157.0,
+      "eval_math_runtime": 35.799,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 3000
+    },
+    {
+      "epoch": 4.8,
+      "eval_physics_entropy": 0.2788742837905884,
+      "eval_physics_loss": 1.0801695585250854,
+      "eval_physics_mean_token_accuracy": 0.7987381658554077,
+      "eval_physics_num_tokens": 85856157.0,
+      "eval_physics_runtime": 41.1763,
+      "eval_physics_samples_per_second": 12.143,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 3000
+    },
+    {
+      "entropy": 0.2044725831132382,
+      "epoch": 4.816,
+      "grad_norm": 2.4375,
+      "learning_rate": 1.553555555555556e-05,
+      "loss": 0.1707,
+      "mean_token_accuracy": 0.9451114404946566,
+      "num_tokens": 86148164.0,
+      "step": 3010
+    },
+    {
+      "entropy": 0.20232565151527523,
+      "epoch": 4.832,
+      "grad_norm": 2.25,
+      "learning_rate": 1.5513333333333333e-05,
+      "loss": 0.1684,
+      "mean_token_accuracy": 0.9456779569387436,
+      "num_tokens": 86439959.0,
+      "step": 3020
+    },
+    {
+      "entropy": 0.20742189674638212,
+      "epoch": 4.848,
+      "grad_norm": 2.3125,
+      "learning_rate": 1.5491111111111114e-05,
+      "loss": 0.172,
+      "mean_token_accuracy": 0.9437137056142092,
+      "num_tokens": 86721776.0,
+      "step": 3030
+    },
+    {
+      "entropy": 0.204735844489187,
+      "epoch": 4.864,
+      "grad_norm": 2.28125,
+      "learning_rate": 1.546888888888889e-05,
+      "loss": 0.168,
+      "mean_token_accuracy": 0.9456948045641184,
+      "num_tokens": 87004064.0,
+      "step": 3040
+    },
+    {
+      "entropy": 0.20204079821705817,
+      "epoch": 4.88,
+      "grad_norm": 2.34375,
+      "learning_rate": 1.544666666666667e-05,
+      "loss": 0.1691,
+      "mean_token_accuracy": 0.9455896012485028,
+      "num_tokens": 87300720.0,
+      "step": 3050
+    },
+    {
+      "entropy": 0.20231212503276766,
+      "epoch": 4.896,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5424444444444447e-05,
+      "loss": 0.1673,
+      "mean_token_accuracy": 0.9458602901548148,
+      "num_tokens": 87593142.0,
+      "step": 3060
+    },
+    {
+      "entropy": 0.19865587293170392,
+      "epoch": 4.912,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.5402222222222224e-05,
+      "loss": 0.1644,
+      "mean_token_accuracy": 0.9465924687683582,
+      "num_tokens": 87881200.0,
+      "step": 3070
+    },
+    {
+      "entropy": 0.201658932864666,
+      "epoch": 4.928,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.5380000000000002e-05,
+      "loss": 0.1669,
+      "mean_token_accuracy": 0.9455430846661329,
+      "num_tokens": 88161276.0,
+      "step": 3080
+    },
+    {
+      "entropy": 0.20509861945174634,
+      "epoch": 4.944,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.535777777777778e-05,
+      "loss": 0.1739,
+      "mean_token_accuracy": 0.9440295215696096,
+      "num_tokens": 88442647.0,
+      "step": 3090
+    },
+    {
+      "entropy": 0.20068702194839716,
+      "epoch": 4.96,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5335555555555557e-05,
+      "loss": 0.1638,
+      "mean_token_accuracy": 0.9464703939855099,
+      "num_tokens": 88727638.0,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_math_entropy": 0.2880592755079269,
+      "eval_math_loss": 0.9388962388038635,
+      "eval_math_mean_token_accuracy": 0.8168627028465271,
+      "eval_math_num_tokens": 88727638.0,
+      "eval_math_runtime": 35.7707,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3100
+    },
+    {
+      "epoch": 4.96,
+      "eval_physics_entropy": 0.2795389975309372,
+      "eval_physics_loss": 1.0800063610076904,
+      "eval_physics_mean_token_accuracy": 0.7986403579711914,
+      "eval_physics_num_tokens": 88727638.0,
+      "eval_physics_runtime": 41.4648,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 3100
+    },
+    {
+      "entropy": 0.20669097322970628,
+      "epoch": 4.976,
+      "grad_norm": 2.40625,
+      "learning_rate": 1.5313333333333335e-05,
+      "loss": 0.1734,
+      "mean_token_accuracy": 0.9433898750692606,
+      "num_tokens": 89014226.0,
+      "step": 3110
+    },
+    {
+      "entropy": 0.20596866989508272,
+      "epoch": 4.992,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.5291111111111112e-05,
+      "loss": 0.1728,
+      "mean_token_accuracy": 0.9439191181212664,
+      "num_tokens": 89291373.0,
+      "step": 3120
+    },
+    {
+      "entropy": 0.17975524491630496,
+      "epoch": 5.008,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.526888888888889e-05,
+      "loss": 0.1294,
+      "mean_token_accuracy": 0.9600633040070534,
+      "num_tokens": 89575899.0,
+      "step": 3130
+    },
+    {
+      "entropy": 0.12737601874396204,
+      "epoch": 5.024,
+      "grad_norm": 1.5234375,
+      "learning_rate": 1.5246666666666668e-05,
+      "loss": 0.0799,
+      "mean_token_accuracy": 0.9776071958243847,
+      "num_tokens": 89860193.0,
+      "step": 3140
+    },
+    {
+      "entropy": 0.12302283146418631,
+      "epoch": 5.04,
+      "grad_norm": 1.546875,
+      "learning_rate": 1.5224444444444447e-05,
+      "loss": 0.0763,
+      "mean_token_accuracy": 0.9784069903194904,
+      "num_tokens": 90155473.0,
+      "step": 3150
+    },
+    {
+      "entropy": 0.1233028472866863,
+      "epoch": 5.056,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5202222222222223e-05,
+      "loss": 0.076,
+      "mean_token_accuracy": 0.9784812565892935,
+      "num_tokens": 90447777.0,
+      "step": 3160
+    },
+    {
+      "entropy": 0.12666487065143883,
+      "epoch": 5.072,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.5180000000000002e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9770117931067943,
+      "num_tokens": 90733638.0,
+      "step": 3170
+    },
+    {
+      "entropy": 0.12284240419976414,
+      "epoch": 5.088,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.5157777777777778e-05,
+      "loss": 0.078,
+      "mean_token_accuracy": 0.9779970541596412,
+      "num_tokens": 91024199.0,
+      "step": 3180
+    },
+    {
+      "entropy": 0.11969875160139054,
+      "epoch": 5.104,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.5135555555555557e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9783815696835518,
+      "num_tokens": 91314842.0,
+      "step": 3190
+    },
+    {
+      "entropy": 0.11972461631521583,
+      "epoch": 5.12,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.5113333333333335e-05,
+      "loss": 0.0758,
+      "mean_token_accuracy": 0.9785178400576114,
+      "num_tokens": 91602437.0,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_math_entropy": 0.24255421686172485,
+      "eval_math_loss": 1.0493830442428589,
+      "eval_math_mean_token_accuracy": 0.8152036366462707,
+      "eval_math_num_tokens": 91602437.0,
+      "eval_math_runtime": 35.752,
+      "eval_math_samples_per_second": 13.985,
+      "eval_math_steps_per_second": 3.496,
+      "step": 3200
+    },
+    {
+      "epoch": 5.12,
+      "eval_physics_entropy": 0.25365390503406526,
+      "eval_physics_loss": 1.161871314048767,
+      "eval_physics_mean_token_accuracy": 0.7974244031906128,
+      "eval_physics_num_tokens": 91602437.0,
+      "eval_physics_runtime": 41.4324,
+      "eval_physics_samples_per_second": 12.068,
+      "eval_physics_steps_per_second": 3.017,
+      "step": 3200
+    },
+    {
+      "entropy": 0.11904104156419634,
+      "epoch": 5.136,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5091111111111113e-05,
+      "loss": 0.0766,
+      "mean_token_accuracy": 0.9780814602971077,
+      "num_tokens": 91887940.0,
+      "step": 3210
+    },
+    {
+      "entropy": 0.12109884731471539,
+      "epoch": 5.152,
+      "grad_norm": 1.6171875,
+      "learning_rate": 1.506888888888889e-05,
+      "loss": 0.0776,
+      "mean_token_accuracy": 0.9777121197432279,
+      "num_tokens": 92169269.0,
+      "step": 3220
+    },
+    {
+      "entropy": 0.11855497197248041,
+      "epoch": 5.168,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.5046666666666668e-05,
+      "loss": 0.0761,
+      "mean_token_accuracy": 0.978103457018733,
+      "num_tokens": 92463326.0,
+      "step": 3230
+    },
+    {
+      "entropy": 0.12662705583497882,
+      "epoch": 5.184,
+      "grad_norm": 1.5390625,
+      "learning_rate": 1.5024444444444445e-05,
+      "loss": 0.081,
+      "mean_token_accuracy": 0.9766223061829805,
+      "num_tokens": 92753062.0,
+      "step": 3240
+    },
+    {
+      "entropy": 0.12166502382606267,
+      "epoch": 5.2,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.5002222222222223e-05,
+      "loss": 0.0773,
+      "mean_token_accuracy": 0.9779473695904016,
+      "num_tokens": 93039536.0,
+      "step": 3250
+    },
+    {
+      "entropy": 0.12320108511485159,
+      "epoch": 5.216,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.498e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9769222058355809,
+      "num_tokens": 93324998.0,
+      "step": 3260
+    },
+    {
+      "entropy": 0.12509258496575057,
+      "epoch": 5.232,
+      "grad_norm": 1.6875,
+      "learning_rate": 1.495777777777778e-05,
+      "loss": 0.0811,
+      "mean_token_accuracy": 0.9771292954683304,
+      "num_tokens": 93607538.0,
+      "step": 3270
+    },
+    {
+      "entropy": 0.12425780035555363,
+      "epoch": 5.248,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4935555555555556e-05,
+      "loss": 0.08,
+      "mean_token_accuracy": 0.976675358787179,
+      "num_tokens": 93889153.0,
+      "step": 3280
+    },
+    {
+      "entropy": 0.12271388149820268,
+      "epoch": 5.264,
+      "grad_norm": 1.5625,
+      "learning_rate": 1.4913333333333335e-05,
+      "loss": 0.0789,
+      "mean_token_accuracy": 0.977228007465601,
+      "num_tokens": 94174451.0,
+      "step": 3290
+    },
+    {
+      "entropy": 0.1203123148996383,
+      "epoch": 5.28,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4891111111111111e-05,
+      "loss": 0.0771,
+      "mean_token_accuracy": 0.9775407303124666,
+      "num_tokens": 94463178.0,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_math_entropy": 0.23942122876644134,
+      "eval_math_loss": 1.0654295682907104,
+      "eval_math_mean_token_accuracy": 0.8145109858512879,
+      "eval_math_num_tokens": 94463178.0,
+      "eval_math_runtime": 35.7653,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3300
+    },
+    {
+      "epoch": 5.28,
+      "eval_physics_entropy": 0.25065979206562045,
+      "eval_physics_loss": 1.1764365434646606,
+      "eval_physics_mean_token_accuracy": 0.7967135186195373,
+      "eval_physics_num_tokens": 94463178.0,
+      "eval_physics_runtime": 41.4451,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3300
+    },
+    {
+      "entropy": 0.12212947504594922,
+      "epoch": 5.296,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.486888888888889e-05,
+      "loss": 0.0797,
+      "mean_token_accuracy": 0.9769365161657333,
+      "num_tokens": 94754637.0,
+      "step": 3310
+    },
+    {
+      "entropy": 0.12625247961841524,
+      "epoch": 5.312,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.4846666666666666e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9761669352650643,
+      "num_tokens": 95034199.0,
+      "step": 3320
+    },
+    {
+      "entropy": 0.12353909476660192,
+      "epoch": 5.328,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4824444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9766261216253043,
+      "num_tokens": 95324901.0,
+      "step": 3330
+    },
+    {
+      "entropy": 0.12188207178842277,
+      "epoch": 5.344,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4802222222222225e-05,
+      "loss": 0.0782,
+      "mean_token_accuracy": 0.9771796323359012,
+      "num_tokens": 95615328.0,
+      "step": 3340
+    },
+    {
+      "entropy": 0.1235438220668584,
+      "epoch": 5.36,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.478e-05,
+      "loss": 0.0818,
+      "mean_token_accuracy": 0.9761600013822317,
+      "num_tokens": 95906341.0,
+      "step": 3350
+    },
+    {
+      "entropy": 0.1280383275821805,
+      "epoch": 5.376,
+      "grad_norm": 1.6328125,
+      "learning_rate": 1.475777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9760391939431429,
+      "num_tokens": 96188769.0,
+      "step": 3360
+    },
+    {
+      "entropy": 0.12412518821656704,
+      "epoch": 5.392,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4735555555555556e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9760548025369644,
+      "num_tokens": 96472047.0,
+      "step": 3370
+    },
+    {
+      "entropy": 0.12685924135148524,
+      "epoch": 5.408,
+      "grad_norm": 1.5546875,
+      "learning_rate": 1.4713333333333335e-05,
+      "loss": 0.0828,
+      "mean_token_accuracy": 0.9758801158517599,
+      "num_tokens": 96752754.0,
+      "step": 3380
+    },
+    {
+      "entropy": 0.12485403921455145,
+      "epoch": 5.424,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4691111111111111e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9760515049099923,
+      "num_tokens": 97042615.0,
+      "step": 3390
+    },
+    {
+      "entropy": 0.12233043625019491,
+      "epoch": 5.44,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.466888888888889e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9764630265533925,
+      "num_tokens": 97324337.0,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_math_entropy": 0.24116922390460968,
+      "eval_math_loss": 1.0628420114517212,
+      "eval_math_mean_token_accuracy": 0.8142255573272705,
+      "eval_math_num_tokens": 97324337.0,
+      "eval_math_runtime": 35.7647,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3400
+    },
+    {
+      "epoch": 5.44,
+      "eval_physics_entropy": 0.2520143152475357,
+      "eval_physics_loss": 1.171416997909546,
+      "eval_physics_mean_token_accuracy": 0.7969081449508667,
+      "eval_physics_num_tokens": 97324337.0,
+      "eval_physics_runtime": 41.7765,
+      "eval_physics_samples_per_second": 11.968,
+      "eval_physics_steps_per_second": 2.992,
+      "step": 3400
+    },
+    {
+      "entropy": 0.12518517016433178,
+      "epoch": 5.456,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4646666666666666e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.975970969721675,
+      "num_tokens": 97610848.0,
+      "step": 3410
+    },
+    {
+      "entropy": 0.12347186738625168,
+      "epoch": 5.4719999999999995,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4624444444444446e-05,
+      "loss": 0.0809,
+      "mean_token_accuracy": 0.9765568733215332,
+      "num_tokens": 97894197.0,
+      "step": 3420
+    },
+    {
+      "entropy": 0.12211509156040848,
+      "epoch": 5.4879999999999995,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4602222222222225e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9765133555978537,
+      "num_tokens": 98178647.0,
+      "step": 3430
+    },
+    {
+      "entropy": 0.1255460273474455,
+      "epoch": 5.504,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.4580000000000001e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.975815936923027,
+      "num_tokens": 98466145.0,
+      "step": 3440
+    },
+    {
+      "entropy": 0.12366767269559205,
+      "epoch": 5.52,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.455777777777778e-05,
+      "loss": 0.0833,
+      "mean_token_accuracy": 0.9755478825420141,
+      "num_tokens": 98747917.0,
+      "step": 3450
+    },
+    {
+      "entropy": 0.1271245769225061,
+      "epoch": 5.536,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4535555555555556e-05,
+      "loss": 0.0854,
+      "mean_token_accuracy": 0.975119874253869,
+      "num_tokens": 99030810.0,
+      "step": 3460
+    },
+    {
+      "entropy": 0.12403408349491656,
+      "epoch": 5.552,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4513333333333336e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762228842824697,
+      "num_tokens": 99318677.0,
+      "step": 3470
+    },
+    {
+      "entropy": 0.1286336882505566,
+      "epoch": 5.568,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.4491111111111111e-05,
+      "loss": 0.084,
+      "mean_token_accuracy": 0.9754358228296042,
+      "num_tokens": 99592537.0,
+      "step": 3480
+    },
+    {
+      "entropy": 0.12519475985318423,
+      "epoch": 5.584,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.446888888888889e-05,
+      "loss": 0.0836,
+      "mean_token_accuracy": 0.9757570996880531,
+      "num_tokens": 99875992.0,
+      "step": 3490
+    },
+    {
+      "entropy": 0.12342443629167973,
+      "epoch": 5.6,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.4446666666666668e-05,
+      "loss": 0.0802,
+      "mean_token_accuracy": 0.9769513830542564,
+      "num_tokens": 100168027.0,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_math_entropy": 0.23975398194789888,
+      "eval_math_loss": 1.071073055267334,
+      "eval_math_mean_token_accuracy": 0.814323438167572,
+      "eval_math_num_tokens": 100168027.0,
+      "eval_math_runtime": 35.7663,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 3500
+    },
+    {
+      "epoch": 5.6,
+      "eval_physics_entropy": 0.2488333419561386,
+      "eval_physics_loss": 1.1844922304153442,
+      "eval_physics_mean_token_accuracy": 0.7968628706932068,
+      "eval_physics_num_tokens": 100168027.0,
+      "eval_physics_runtime": 41.1644,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 3500
+    },
+    {
+      "entropy": 0.11995140444487333,
+      "epoch": 5.616,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4424444444444446e-05,
+      "loss": 0.0804,
+      "mean_token_accuracy": 0.976571949198842,
+      "num_tokens": 100458800.0,
+      "step": 3510
+    },
+    {
+      "entropy": 0.12303966791369021,
+      "epoch": 5.632,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4402222222222224e-05,
+      "loss": 0.083,
+      "mean_token_accuracy": 0.9757887721061707,
+      "num_tokens": 100750138.0,
+      "step": 3520
+    },
+    {
+      "entropy": 0.12365907938219607,
+      "epoch": 5.648,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9761408776044845,
+      "num_tokens": 101035408.0,
+      "step": 3530
+    },
+    {
+      "entropy": 0.12068781261332333,
+      "epoch": 5.664,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4357777777777779e-05,
+      "loss": 0.0801,
+      "mean_token_accuracy": 0.9766639798879624,
+      "num_tokens": 101331186.0,
+      "step": 3540
+    },
+    {
+      "entropy": 0.12283993698656559,
+      "epoch": 5.68,
+      "grad_norm": 1.9609375,
+      "learning_rate": 1.4335555555555556e-05,
+      "loss": 0.0805,
+      "mean_token_accuracy": 0.9763456650078297,
+      "num_tokens": 101621993.0,
+      "step": 3550
+    },
+    {
+      "entropy": 0.12474635057151318,
+      "epoch": 5.696,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.4313333333333334e-05,
+      "loss": 0.0825,
+      "mean_token_accuracy": 0.9756041605025529,
+      "num_tokens": 101904112.0,
+      "step": 3560
+    },
+    {
+      "entropy": 0.12090231254696845,
+      "epoch": 5.712,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.4291111111111113e-05,
+      "loss": 0.082,
+      "mean_token_accuracy": 0.9761635400354862,
+      "num_tokens": 102196831.0,
+      "step": 3570
+    },
+    {
+      "entropy": 0.12512639928609132,
+      "epoch": 5.728,
+      "grad_norm": 2.0,
+      "learning_rate": 1.426888888888889e-05,
+      "loss": 0.0839,
+      "mean_token_accuracy": 0.9757933501154185,
+      "num_tokens": 102482122.0,
+      "step": 3580
+    },
+    {
+      "entropy": 0.12669197719078512,
+      "epoch": 5.744,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.4246666666666669e-05,
+      "loss": 0.0845,
+      "mean_token_accuracy": 0.9752174407243729,
+      "num_tokens": 102754317.0,
+      "step": 3590
+    },
+    {
+      "entropy": 0.12015235424041748,
+      "epoch": 5.76,
+      "grad_norm": 1.6015625,
+      "learning_rate": 1.4224444444444445e-05,
+      "loss": 0.0795,
+      "mean_token_accuracy": 0.9764200799167156,
+      "num_tokens": 103040247.0,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_math_entropy": 0.23906498980522156,
+      "eval_math_loss": 1.0675584077835083,
+      "eval_math_mean_token_accuracy": 0.8141876225471496,
+      "eval_math_num_tokens": 103040247.0,
+      "eval_math_runtime": 35.8224,
+      "eval_math_samples_per_second": 13.958,
+      "eval_math_steps_per_second": 3.489,
+      "step": 3600
+    },
+    {
+      "epoch": 5.76,
+      "eval_physics_entropy": 0.24975030255317687,
+      "eval_physics_loss": 1.178022027015686,
+      "eval_physics_mean_token_accuracy": 0.7968517804145813,
+      "eval_physics_num_tokens": 103040247.0,
+      "eval_physics_runtime": 41.4817,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 3600
+    },
+    {
+      "entropy": 0.12267617424950003,
+      "epoch": 5.776,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.4202222222222224e-05,
+      "loss": 0.0814,
+      "mean_token_accuracy": 0.9759464427828789,
+      "num_tokens": 103322676.0,
+      "step": 3610
+    },
+    {
+      "entropy": 0.12787257870659233,
+      "epoch": 5.792,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.418e-05,
+      "loss": 0.0866,
+      "mean_token_accuracy": 0.9747348535805941,
+      "num_tokens": 103601932.0,
+      "step": 3620
+    },
+    {
+      "entropy": 0.12417071931995452,
+      "epoch": 5.808,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.4157777777777779e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9756274610757828,
+      "num_tokens": 103899790.0,
+      "step": 3630
+    },
+    {
+      "entropy": 0.12282869541086257,
+      "epoch": 5.824,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4135555555555555e-05,
+      "loss": 0.0815,
+      "mean_token_accuracy": 0.9763524547219277,
+      "num_tokens": 104178127.0,
+      "step": 3640
+    },
+    {
+      "entropy": 0.12581051201559604,
+      "epoch": 5.84,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.4113333333333334e-05,
+      "loss": 0.0853,
+      "mean_token_accuracy": 0.9752345487475396,
+      "num_tokens": 104468018.0,
+      "step": 3650
+    },
+    {
+      "entropy": 0.12290314708370716,
+      "epoch": 5.856,
+      "grad_norm": 2.0,
+      "learning_rate": 1.4091111111111114e-05,
+      "loss": 0.0824,
+      "mean_token_accuracy": 0.9761314608156682,
+      "num_tokens": 104760904.0,
+      "step": 3660
+    },
+    {
+      "entropy": 0.12341588027775288,
+      "epoch": 5.872,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.406888888888889e-05,
+      "loss": 0.0819,
+      "mean_token_accuracy": 0.9755884770303964,
+      "num_tokens": 105040296.0,
+      "step": 3670
+    },
+    {
+      "entropy": 0.12345146541483701,
+      "epoch": 5.888,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4046666666666669e-05,
+      "loss": 0.0829,
+      "mean_token_accuracy": 0.9762541994452476,
+      "num_tokens": 105334020.0,
+      "step": 3680
+    },
+    {
+      "entropy": 0.12768929479643704,
+      "epoch": 5.904,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4024444444444445e-05,
+      "loss": 0.0844,
+      "mean_token_accuracy": 0.9747776690870523,
+      "num_tokens": 105618594.0,
+      "step": 3690
+    },
+    {
+      "entropy": 0.1248483968898654,
+      "epoch": 5.92,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.4002222222222224e-05,
+      "loss": 0.0842,
+      "mean_token_accuracy": 0.9751092988997698,
+      "num_tokens": 105908769.0,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_math_entropy": 0.23840607726573945,
+      "eval_math_loss": 1.0716767311096191,
+      "eval_math_mean_token_accuracy": 0.8149038062095643,
+      "eval_math_num_tokens": 105908769.0,
+      "eval_math_runtime": 35.7392,
+      "eval_math_samples_per_second": 13.99,
+      "eval_math_steps_per_second": 3.498,
+      "step": 3700
+    },
+    {
+      "epoch": 5.92,
+      "eval_physics_entropy": 0.24841285252571105,
+      "eval_physics_loss": 1.184152603149414,
+      "eval_physics_mean_token_accuracy": 0.7970950255393981,
+      "eval_physics_num_tokens": 105908769.0,
+      "eval_physics_runtime": 41.4426,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3700
+    },
+    {
+      "entropy": 0.12872918481007217,
+      "epoch": 5.936,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.398e-05,
+      "loss": 0.0859,
+      "mean_token_accuracy": 0.9748044356703758,
+      "num_tokens": 106184251.0,
+      "step": 3710
+    },
+    {
+      "entropy": 0.12503819265402855,
+      "epoch": 5.952,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.395777777777778e-05,
+      "loss": 0.0827,
+      "mean_token_accuracy": 0.9756678476929664,
+      "num_tokens": 106475238.0,
+      "step": 3720
+    },
+    {
+      "entropy": 0.12310672523453832,
+      "epoch": 5.968,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3935555555555557e-05,
+      "loss": 0.0832,
+      "mean_token_accuracy": 0.9751558441668748,
+      "num_tokens": 106761563.0,
+      "step": 3730
+    },
+    {
+      "entropy": 0.12658624798059465,
+      "epoch": 5.984,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.3913333333333335e-05,
+      "loss": 0.0848,
+      "mean_token_accuracy": 0.9749173305928707,
+      "num_tokens": 107044983.0,
+      "step": 3740
+    },
+    {
+      "entropy": 0.1277452927082777,
+      "epoch": 6.0,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.3891111111111114e-05,
+      "loss": 0.0857,
+      "mean_token_accuracy": 0.9748894535005093,
+      "num_tokens": 107318544.0,
+      "step": 3750
+    },
+    {
+      "entropy": 0.08838868229649961,
+      "epoch": 6.016,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.386888888888889e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9871940586715937,
+      "num_tokens": 107597955.0,
+      "step": 3760
+    },
+    {
+      "entropy": 0.08257517793681472,
+      "epoch": 6.032,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3846666666666669e-05,
+      "loss": 0.0478,
+      "mean_token_accuracy": 0.9871934026479721,
+      "num_tokens": 107882792.0,
+      "step": 3770
+    },
+    {
+      "entropy": 0.0828174332389608,
+      "epoch": 6.048,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3824444444444445e-05,
+      "loss": 0.0465,
+      "mean_token_accuracy": 0.9873949829488993,
+      "num_tokens": 108172211.0,
+      "step": 3780
+    },
+    {
+      "entropy": 0.0837429367704317,
+      "epoch": 6.064,
+      "grad_norm": 1.3671875,
+      "learning_rate": 1.3802222222222224e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9872307199984789,
+      "num_tokens": 108456311.0,
+      "step": 3790
+    },
+    {
+      "entropy": 0.08258170303888619,
+      "epoch": 6.08,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.378e-05,
+      "loss": 0.0469,
+      "mean_token_accuracy": 0.9875688355416059,
+      "num_tokens": 108741840.0,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_math_entropy": 0.2157260871529579,
+      "eval_math_loss": 1.1480746269226074,
+      "eval_math_mean_token_accuracy": 0.8150639281272888,
+      "eval_math_num_tokens": 108741840.0,
+      "eval_math_runtime": 35.7792,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 3800
+    },
+    {
+      "epoch": 6.08,
+      "eval_physics_entropy": 0.2341501749753952,
+      "eval_physics_loss": 1.242053747177124,
+      "eval_physics_mean_token_accuracy": 0.7970363121032715,
+      "eval_physics_num_tokens": 108741840.0,
+      "eval_physics_runtime": 41.45,
+      "eval_physics_samples_per_second": 12.063,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3800
+    },
+    {
+      "entropy": 0.08380906586535275,
+      "epoch": 6.096,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.375777777777778e-05,
+      "loss": 0.0485,
+      "mean_token_accuracy": 0.9867967016994953,
+      "num_tokens": 109019117.0,
+      "step": 3810
+    },
+    {
+      "entropy": 0.08307188230101019,
+      "epoch": 6.112,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3735555555555557e-05,
+      "loss": 0.048,
+      "mean_token_accuracy": 0.9867318440228701,
+      "num_tokens": 109311190.0,
+      "step": 3820
+    },
+    {
+      "entropy": 0.08322459494229406,
+      "epoch": 6.128,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3713333333333335e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.986859867721796,
+      "num_tokens": 109597456.0,
+      "step": 3830
+    },
+    {
+      "entropy": 0.0813226085389033,
+      "epoch": 6.144,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3691111111111112e-05,
+      "loss": 0.0473,
+      "mean_token_accuracy": 0.9870293717831373,
+      "num_tokens": 109889318.0,
+      "step": 3840
+    },
+    {
+      "entropy": 0.08239689606707543,
+      "epoch": 6.16,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.366888888888889e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9869503367692232,
+      "num_tokens": 110172819.0,
+      "step": 3850
+    },
+    {
+      "entropy": 0.08577790288254619,
+      "epoch": 6.176,
+      "grad_norm": 1.3828125,
+      "learning_rate": 1.3646666666666668e-05,
+      "loss": 0.0492,
+      "mean_token_accuracy": 0.9867170631885529,
+      "num_tokens": 110459481.0,
+      "step": 3860
+    },
+    {
+      "entropy": 0.08229333041235805,
+      "epoch": 6.192,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3624444444444445e-05,
+      "loss": 0.049,
+      "mean_token_accuracy": 0.9867132067680359,
+      "num_tokens": 110747009.0,
+      "step": 3870
+    },
+    {
+      "entropy": 0.08104511541314423,
+      "epoch": 6.208,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.3602222222222223e-05,
+      "loss": 0.0477,
+      "mean_token_accuracy": 0.9871339503675699,
+      "num_tokens": 111034380.0,
+      "step": 3880
+    },
+    {
+      "entropy": 0.08298772852867842,
+      "epoch": 6.224,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3580000000000002e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9868796184659004,
+      "num_tokens": 111326385.0,
+      "step": 3890
+    },
+    {
+      "entropy": 0.08269706610590219,
+      "epoch": 6.24,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3557777777777778e-05,
+      "loss": 0.0481,
+      "mean_token_accuracy": 0.9872248943895101,
+      "num_tokens": 111617177.0,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_math_entropy": 0.2116558358669281,
+      "eval_math_loss": 1.1540143489837646,
+      "eval_math_mean_token_accuracy": 0.8150960731506348,
+      "eval_math_num_tokens": 111617177.0,
+      "eval_math_runtime": 35.7843,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 3900
+    },
+    {
+      "epoch": 6.24,
+      "eval_physics_entropy": 0.23068565714359285,
+      "eval_physics_loss": 1.244989275932312,
+      "eval_physics_mean_token_accuracy": 0.797289421081543,
+      "eval_physics_num_tokens": 111617177.0,
+      "eval_physics_runtime": 41.443,
+      "eval_physics_samples_per_second": 12.065,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 3900
+    },
+    {
+      "entropy": 0.08489166686777025,
+      "epoch": 6.256,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.3535555555555557e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9859484441578388,
+      "num_tokens": 111893281.0,
+      "step": 3910
+    },
+    {
+      "entropy": 0.08082030359655619,
+      "epoch": 6.272,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3513333333333333e-05,
+      "loss": 0.0476,
+      "mean_token_accuracy": 0.9871520031243562,
+      "num_tokens": 112189253.0,
+      "step": 3920
+    },
+    {
+      "entropy": 0.08449386192951351,
+      "epoch": 6.288,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3491111111111113e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986621393635869,
+      "num_tokens": 112473705.0,
+      "step": 3930
+    },
+    {
+      "entropy": 0.08417739674914629,
+      "epoch": 6.304,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.3468888888888888e-05,
+      "loss": 0.0503,
+      "mean_token_accuracy": 0.9866305027157068,
+      "num_tokens": 112753183.0,
+      "step": 3940
+    },
+    {
+      "entropy": 0.08426821974571794,
+      "epoch": 6.32,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3446666666666668e-05,
+      "loss": 0.0496,
+      "mean_token_accuracy": 0.9866181764751673,
+      "num_tokens": 113040028.0,
+      "step": 3950
+    },
+    {
+      "entropy": 0.08344817301258445,
+      "epoch": 6.336,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3424444444444447e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863654665648938,
+      "num_tokens": 113323947.0,
+      "step": 3960
+    },
+    {
+      "entropy": 0.08314311665017157,
+      "epoch": 6.352,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.3402222222222223e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9866466347128153,
+      "num_tokens": 113603514.0,
+      "step": 3970
+    },
+    {
+      "entropy": 0.08517711046151817,
+      "epoch": 6.368,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.3380000000000002e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9866021793335676,
+      "num_tokens": 113894528.0,
+      "step": 3980
+    },
+    {
+      "entropy": 0.08286653154063969,
+      "epoch": 6.384,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3357777777777778e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.986630416661501,
+      "num_tokens": 114172836.0,
+      "step": 3990
+    },
+    {
+      "entropy": 0.082943707681261,
+      "epoch": 6.4,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.3335555555555558e-05,
+      "loss": 0.0491,
+      "mean_token_accuracy": 0.9870663437992334,
+      "num_tokens": 114460847.0,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_math_entropy": 0.21333465337753296,
+      "eval_math_loss": 1.1600124835968018,
+      "eval_math_mean_token_accuracy": 0.815093065738678,
+      "eval_math_num_tokens": 114460847.0,
+      "eval_math_runtime": 35.7657,
+      "eval_math_samples_per_second": 13.98,
+      "eval_math_steps_per_second": 3.495,
+      "step": 4000
+    },
+    {
+      "epoch": 6.4,
+      "eval_physics_entropy": 0.23318842422962188,
+      "eval_physics_loss": 1.2491592168807983,
+      "eval_physics_mean_token_accuracy": 0.7966313948631286,
+      "eval_physics_num_tokens": 114460847.0,
+      "eval_physics_runtime": 41.22,
+      "eval_physics_samples_per_second": 12.13,
+      "eval_physics_steps_per_second": 3.033,
+      "step": 4000
+    },
+    {
+      "entropy": 0.08455195005517453,
+      "epoch": 6.416,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3313333333333333e-05,
+      "loss": 0.0504,
+      "mean_token_accuracy": 0.9863806873559952,
+      "num_tokens": 114744933.0,
+      "step": 4010
+    },
+    {
+      "entropy": 0.08246206922922283,
+      "epoch": 6.432,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.3291111111111113e-05,
+      "loss": 0.0486,
+      "mean_token_accuracy": 0.9869922034442424,
+      "num_tokens": 115035799.0,
+      "step": 4020
+    },
+    {
+      "entropy": 0.08484494448639453,
+      "epoch": 6.448,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3268888888888889e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9866210386157036,
+      "num_tokens": 115324970.0,
+      "step": 4030
+    },
+    {
+      "entropy": 0.08431054749526083,
+      "epoch": 6.464,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.3246666666666668e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865565162152052,
+      "num_tokens": 115617802.0,
+      "step": 4040
+    },
+    {
+      "entropy": 0.0853661192348227,
+      "epoch": 6.48,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.3224444444444446e-05,
+      "loss": 0.051,
+      "mean_token_accuracy": 0.9861403401941061,
+      "num_tokens": 115895920.0,
+      "step": 4050
+    },
+    {
+      "entropy": 0.0857542624231428,
+      "epoch": 6.496,
+      "grad_norm": 1.21875,
+      "learning_rate": 1.3202222222222223e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9863781563937664,
+      "num_tokens": 116183103.0,
+      "step": 4060
+    },
+    {
+      "entropy": 0.08211216740310193,
+      "epoch": 6.5120000000000005,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0489,
+      "mean_token_accuracy": 0.9870047964155674,
+      "num_tokens": 116470867.0,
+      "step": 4070
+    },
+    {
+      "entropy": 0.08192372117191553,
+      "epoch": 6.5280000000000005,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.3157777777777778e-05,
+      "loss": 0.0488,
+      "mean_token_accuracy": 0.9866585023701191,
+      "num_tokens": 116761140.0,
+      "step": 4080
+    },
+    {
+      "entropy": 0.08760172158945352,
+      "epoch": 6.5440000000000005,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.3135555555555558e-05,
+      "loss": 0.0513,
+      "mean_token_accuracy": 0.9858183633536101,
+      "num_tokens": 117044558.0,
+      "step": 4090
+    },
+    {
+      "entropy": 0.0826570155331865,
+      "epoch": 6.5600000000000005,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3113333333333334e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865829836577177,
+      "num_tokens": 117338304.0,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_math_entropy": 0.21119803923368455,
+      "eval_math_loss": 1.1628667116165161,
+      "eval_math_mean_token_accuracy": 0.8154761862754821,
+      "eval_math_num_tokens": 117338304.0,
+      "eval_math_runtime": 35.8268,
+      "eval_math_samples_per_second": 13.956,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4100
+    },
+    {
+      "epoch": 6.5600000000000005,
+      "eval_physics_entropy": 0.22983593618869783,
+      "eval_physics_loss": 1.2532014846801758,
+      "eval_physics_mean_token_accuracy": 0.7969202942848206,
+      "eval_physics_num_tokens": 117338304.0,
+      "eval_physics_runtime": 41.5321,
+      "eval_physics_samples_per_second": 12.039,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 4100
+    },
+    {
+      "entropy": 0.08333607679232954,
+      "epoch": 6.576,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3091111111111113e-05,
+      "loss": 0.0497,
+      "mean_token_accuracy": 0.9862550985068083,
+      "num_tokens": 117633007.0,
+      "step": 4110
+    },
+    {
+      "entropy": 0.08117786860093475,
+      "epoch": 6.592,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.306888888888889e-05,
+      "loss": 0.0494,
+      "mean_token_accuracy": 0.9866597782820463,
+      "num_tokens": 117927414.0,
+      "step": 4120
+    },
+    {
+      "entropy": 0.08642549277283251,
+      "epoch": 6.608,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.3046666666666668e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9859737418591976,
+      "num_tokens": 118209455.0,
+      "step": 4130
+    },
+    {
+      "entropy": 0.08443117390852421,
+      "epoch": 6.624,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.3024444444444446e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862124625593424,
+      "num_tokens": 118495992.0,
+      "step": 4140
+    },
+    {
+      "entropy": 0.08461201293393969,
+      "epoch": 6.64,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.3002222222222223e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9858912594616414,
+      "num_tokens": 118779177.0,
+      "step": 4150
+    },
+    {
+      "entropy": 0.08362286686897277,
+      "epoch": 6.656,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9861939534544945,
+      "num_tokens": 119065831.0,
+      "step": 4160
+    },
+    {
+      "entropy": 0.08377992056775838,
+      "epoch": 6.672,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2957777777777779e-05,
+      "loss": 0.0505,
+      "mean_token_accuracy": 0.9862146992236376,
+      "num_tokens": 119349974.0,
+      "step": 4170
+    },
+    {
+      "entropy": 0.08517354882787913,
+      "epoch": 6.688,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2935555555555556e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9860025811940432,
+      "num_tokens": 119634067.0,
+      "step": 4180
+    },
+    {
+      "entropy": 0.08468333540949971,
+      "epoch": 6.704,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.2913333333333336e-05,
+      "loss": 0.0506,
+      "mean_token_accuracy": 0.9862395841628313,
+      "num_tokens": 119922640.0,
+      "step": 4190
+    },
+    {
+      "entropy": 0.08581791603937745,
+      "epoch": 6.72,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2891111111111112e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9857941895723343,
+      "num_tokens": 120206747.0,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_math_entropy": 0.2110741074681282,
+      "eval_math_loss": 1.1673368215560913,
+      "eval_math_mean_token_accuracy": 0.8153284401893616,
+      "eval_math_num_tokens": 120206747.0,
+      "eval_math_runtime": 35.8248,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 4200
+    },
+    {
+      "epoch": 6.72,
+      "eval_physics_entropy": 0.22907479107379913,
+      "eval_physics_loss": 1.260624885559082,
+      "eval_physics_mean_token_accuracy": 0.7967779293060303,
+      "eval_physics_num_tokens": 120206747.0,
+      "eval_physics_runtime": 41.5156,
+      "eval_physics_samples_per_second": 12.044,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4200
+    },
+    {
+      "entropy": 0.08579737725667655,
+      "epoch": 6.736,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.2868888888888891e-05,
+      "loss": 0.0516,
+      "mean_token_accuracy": 0.9860372472554445,
+      "num_tokens": 120486581.0,
+      "step": 4210
+    },
+    {
+      "entropy": 0.08400838901288807,
+      "epoch": 6.752,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2846666666666667e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9863463308662176,
+      "num_tokens": 120779617.0,
+      "step": 4220
+    },
+    {
+      "entropy": 0.08478187646251172,
+      "epoch": 6.768,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.2824444444444446e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.9862184945493937,
+      "num_tokens": 121064384.0,
+      "step": 4230
+    },
+    {
+      "entropy": 0.08424062917474658,
+      "epoch": 6.784,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2802222222222222e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9859384100884199,
+      "num_tokens": 121351299.0,
+      "step": 4240
+    },
+    {
+      "entropy": 0.08782856916077435,
+      "epoch": 6.8,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2780000000000001e-05,
+      "loss": 0.0531,
+      "mean_token_accuracy": 0.9860101863741875,
+      "num_tokens": 121626812.0,
+      "step": 4250
+    },
+    {
+      "entropy": 0.08583347962703555,
+      "epoch": 6.816,
+      "grad_norm": 1.3203125,
+      "learning_rate": 1.2757777777777777e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9861073594540357,
+      "num_tokens": 121917563.0,
+      "step": 4260
+    },
+    {
+      "entropy": 0.08525952904019504,
+      "epoch": 6.832,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.2735555555555557e-05,
+      "loss": 0.0509,
+      "mean_token_accuracy": 0.985946049168706,
+      "num_tokens": 122205507.0,
+      "step": 4270
+    },
+    {
+      "entropy": 0.08095580453518778,
+      "epoch": 6.848,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.2713333333333336e-05,
+      "loss": 0.0479,
+      "mean_token_accuracy": 0.9870608881115913,
+      "num_tokens": 122502960.0,
+      "step": 4280
+    },
+    {
+      "entropy": 0.0856839137384668,
+      "epoch": 6.864,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.2691111111111112e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9857846073806286,
+      "num_tokens": 122783468.0,
+      "step": 4290
+    },
+    {
+      "entropy": 0.08315675526391715,
+      "epoch": 6.88,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2668888888888891e-05,
+      "loss": 0.0498,
+      "mean_token_accuracy": 0.9864332474768162,
+      "num_tokens": 123073078.0,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_math_entropy": 0.2092110061645508,
+      "eval_math_loss": 1.1747575998306274,
+      "eval_math_mean_token_accuracy": 0.8151103315353394,
+      "eval_math_num_tokens": 123073078.0,
+      "eval_math_runtime": 35.8349,
+      "eval_math_samples_per_second": 13.953,
+      "eval_math_steps_per_second": 3.488,
+      "step": 4300
+    },
+    {
+      "epoch": 6.88,
+      "eval_physics_entropy": 0.22831890201568603,
+      "eval_physics_loss": 1.264940619468689,
+      "eval_physics_mean_token_accuracy": 0.7968414192199706,
+      "eval_physics_num_tokens": 123073078.0,
+      "eval_physics_runtime": 41.5084,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 4300
+    },
+    {
+      "entropy": 0.08344000696670265,
+      "epoch": 6.896,
+      "grad_norm": 1.2109375,
+      "learning_rate": 1.2646666666666667e-05,
+      "loss": 0.0501,
+      "mean_token_accuracy": 0.9864881403744221,
+      "num_tokens": 123362103.0,
+      "step": 4310
+    },
+    {
+      "entropy": 0.08804852021858096,
+      "epoch": 6.912,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.2624444444444446e-05,
+      "loss": 0.0534,
+      "mean_token_accuracy": 0.9851112041622401,
+      "num_tokens": 123640151.0,
+      "step": 4320
+    },
+    {
+      "entropy": 0.08389839944429696,
+      "epoch": 6.928,
+      "grad_norm": 1.3125,
+      "learning_rate": 1.2602222222222222e-05,
+      "loss": 0.05,
+      "mean_token_accuracy": 0.9865112222731114,
+      "num_tokens": 123930914.0,
+      "step": 4330
+    },
+    {
+      "entropy": 0.08494017843622714,
+      "epoch": 6.944,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.2580000000000002e-05,
+      "loss": 0.0512,
+      "mean_token_accuracy": 0.9860570751130581,
+      "num_tokens": 124216965.0,
+      "step": 4340
+    },
+    {
+      "entropy": 0.08553386291023343,
+      "epoch": 6.96,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2557777777777779e-05,
+      "loss": 0.0517,
+      "mean_token_accuracy": 0.9858683951199054,
+      "num_tokens": 124495374.0,
+      "step": 4350
+    },
+    {
+      "entropy": 0.0846124368486926,
+      "epoch": 6.976,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2535555555555557e-05,
+      "loss": 0.0507,
+      "mean_token_accuracy": 0.9861722022294999,
+      "num_tokens": 124779153.0,
+      "step": 4360
+    },
+    {
+      "entropy": 0.08296751752495765,
+      "epoch": 6.992,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.2513333333333334e-05,
+      "loss": 0.0502,
+      "mean_token_accuracy": 0.9865076046437025,
+      "num_tokens": 125065811.0,
+      "step": 4370
+    },
+    {
+      "entropy": 0.0782727177022025,
+      "epoch": 7.008,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2491111111111112e-05,
+      "loss": 0.0455,
+      "mean_token_accuracy": 0.987744751945138,
+      "num_tokens": 125343099.0,
+      "step": 4380
+    },
+    {
+      "entropy": 0.06690982989966869,
+      "epoch": 7.024,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.246888888888889e-05,
+      "loss": 0.0414,
+      "mean_token_accuracy": 0.9888394929468631,
+      "num_tokens": 125628503.0,
+      "step": 4390
+    },
+    {
+      "entropy": 0.06452394074294716,
+      "epoch": 7.04,
+      "grad_norm": 0.75,
+      "learning_rate": 1.2446666666666667e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892929658293724,
+      "num_tokens": 125916477.0,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_math_entropy": 0.19902950882911682,
+      "eval_math_loss": 1.2132738828659058,
+      "eval_math_mean_token_accuracy": 0.8153792238235473,
+      "eval_math_num_tokens": 125916477.0,
+      "eval_math_runtime": 35.8544,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 4400
+    },
+    {
+      "epoch": 7.04,
+      "eval_physics_entropy": 0.22179586195945739,
+      "eval_physics_loss": 1.2990820407867432,
+      "eval_physics_mean_token_accuracy": 0.7966229004859924,
+      "eval_physics_num_tokens": 125916477.0,
+      "eval_physics_runtime": 41.5422,
+      "eval_physics_samples_per_second": 12.036,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 4400
+    },
+    {
+      "entropy": 0.06385192952584476,
+      "epoch": 7.056,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.2424444444444445e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.9894341301172972,
+      "num_tokens": 126207007.0,
+      "step": 4410
+    },
+    {
+      "entropy": 0.06478747753426432,
+      "epoch": 7.072,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.2402222222222222e-05,
+      "loss": 0.0378,
+      "mean_token_accuracy": 0.9893794670701027,
+      "num_tokens": 126498674.0,
+      "step": 4420
+    },
+    {
+      "entropy": 0.06380244307219982,
+      "epoch": 7.088,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.2380000000000002e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894326873123646,
+      "num_tokens": 126789042.0,
+      "step": 4430
+    },
+    {
+      "entropy": 0.06424271930009126,
+      "epoch": 7.104,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.235777777777778e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9894317038357258,
+      "num_tokens": 127078313.0,
+      "step": 4440
+    },
+    {
+      "entropy": 0.06499261818826199,
+      "epoch": 7.12,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2335555555555557e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892727185040713,
+      "num_tokens": 127363541.0,
+      "step": 4450
+    },
+    {
+      "entropy": 0.06562398292589933,
+      "epoch": 7.136,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.2313333333333335e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9892641551792621,
+      "num_tokens": 127642777.0,
+      "step": 4460
+    },
+    {
+      "entropy": 0.06509807226248085,
+      "epoch": 7.152,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.2291111111111112e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890487994998693,
+      "num_tokens": 127928479.0,
+      "step": 4470
+    },
+    {
+      "entropy": 0.06522750512231142,
+      "epoch": 7.168,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.226888888888889e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.989065657928586,
+      "num_tokens": 128209825.0,
+      "step": 4480
+    },
+    {
+      "entropy": 0.06474495439324528,
+      "epoch": 7.184,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.2246666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9894650772213935,
+      "num_tokens": 128493833.0,
+      "step": 4490
+    },
+    {
+      "entropy": 0.06559755264315754,
+      "epoch": 7.2,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.2224444444444445e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890734907239676,
+      "num_tokens": 128773901.0,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_math_entropy": 0.1962946730852127,
+      "eval_math_loss": 1.2238725423812866,
+      "eval_math_mean_token_accuracy": 0.8156712355613709,
+      "eval_math_num_tokens": 128773901.0,
+      "eval_math_runtime": 35.8136,
+      "eval_math_samples_per_second": 13.961,
+      "eval_math_steps_per_second": 3.49,
+      "step": 4500
+    },
+    {
+      "epoch": 7.2,
+      "eval_physics_entropy": 0.2198963850736618,
+      "eval_physics_loss": 1.3048741817474365,
+      "eval_physics_mean_token_accuracy": 0.7971375894546509,
+      "eval_physics_num_tokens": 128773901.0,
+      "eval_physics_runtime": 41.1717,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 4500
+    },
+    {
+      "entropy": 0.06589861372485757,
+      "epoch": 7.216,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.2202222222222224e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.989148149639368,
+      "num_tokens": 129057801.0,
+      "step": 4510
+    },
+    {
+      "entropy": 0.06478882993105799,
+      "epoch": 7.232,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.218e-05,
+      "loss": 0.0386,
+      "mean_token_accuracy": 0.989036962017417,
+      "num_tokens": 129340781.0,
+      "step": 4520
+    },
+    {
+      "entropy": 0.06308786221779883,
+      "epoch": 7.248,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.215777777777778e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9893654063344002,
+      "num_tokens": 129631132.0,
+      "step": 4530
+    },
+    {
+      "entropy": 0.06536523974500597,
+      "epoch": 7.264,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.2135555555555556e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9890879034996033,
+      "num_tokens": 129913307.0,
+      "step": 4540
+    },
+    {
+      "entropy": 0.06405791421420873,
+      "epoch": 7.28,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.2113333333333335e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891516473144293,
+      "num_tokens": 130201744.0,
+      "step": 4550
+    },
+    {
+      "entropy": 0.06376761447172612,
+      "epoch": 7.296,
+      "grad_norm": 0.75,
+      "learning_rate": 1.209111111111111e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9893031839281321,
+      "num_tokens": 130488187.0,
+      "step": 4560
+    },
+    {
+      "entropy": 0.0670982622075826,
+      "epoch": 7.312,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.206888888888889e-05,
+      "loss": 0.0399,
+      "mean_token_accuracy": 0.9886458057910203,
+      "num_tokens": 130761481.0,
+      "step": 4570
+    },
+    {
+      "entropy": 0.06522584960330277,
+      "epoch": 7.328,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.204666666666667e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9892189968377352,
+      "num_tokens": 131046136.0,
+      "step": 4580
+    },
+    {
+      "entropy": 0.064781365939416,
+      "epoch": 7.344,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.2024444444444445e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9891346648335457,
+      "num_tokens": 131333800.0,
+      "step": 4590
+    },
+    {
+      "entropy": 0.06483790301717818,
+      "epoch": 7.36,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.2002222222222225e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.989232637360692,
+      "num_tokens": 131626450.0,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_math_entropy": 0.19576846724748612,
+      "eval_math_loss": 1.226768136024475,
+      "eval_math_mean_token_accuracy": 0.8159392666816712,
+      "eval_math_num_tokens": 131626450.0,
+      "eval_math_runtime": 35.7948,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4600
+    },
+    {
+      "epoch": 7.36,
+      "eval_physics_entropy": 0.21936397910118102,
+      "eval_physics_loss": 1.3070223331451416,
+      "eval_physics_mean_token_accuracy": 0.7969147138595581,
+      "eval_physics_num_tokens": 131626450.0,
+      "eval_physics_runtime": 41.4955,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 4600
+    },
+    {
+      "entropy": 0.06515917279757559,
+      "epoch": 7.376,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0392,
+      "mean_token_accuracy": 0.9892047554254532,
+      "num_tokens": 131906649.0,
+      "step": 4610
+    },
+    {
+      "entropy": 0.06519709336571396,
+      "epoch": 7.392,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.195777777777778e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9889419995248317,
+      "num_tokens": 132190328.0,
+      "step": 4620
+    },
+    {
+      "entropy": 0.06457304577343166,
+      "epoch": 7.408,
+      "grad_norm": 0.58984375,
+      "learning_rate": 1.1935555555555556e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893009889870882,
+      "num_tokens": 132473728.0,
+      "step": 4630
+    },
+    {
+      "entropy": 0.06410650881007314,
+      "epoch": 7.424,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.1913333333333335e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9890226166695356,
+      "num_tokens": 132758657.0,
+      "step": 4640
+    },
+    {
+      "entropy": 0.0660351888742298,
+      "epoch": 7.44,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1891111111111111e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889182843267917,
+      "num_tokens": 133037748.0,
+      "step": 4650
+    },
+    {
+      "entropy": 0.06468923375941812,
+      "epoch": 7.456,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.186888888888889e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9891764044761657,
+      "num_tokens": 133326761.0,
+      "step": 4660
+    },
+    {
+      "entropy": 0.06298574220854788,
+      "epoch": 7.4719999999999995,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1846666666666668e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.989344285428524,
+      "num_tokens": 133617018.0,
+      "step": 4670
+    },
+    {
+      "entropy": 0.06321970208082348,
+      "epoch": 7.4879999999999995,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1824444444444445e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892230208963155,
+      "num_tokens": 133904602.0,
+      "step": 4680
+    },
+    {
+      "entropy": 0.06326428384054453,
+      "epoch": 7.504,
+      "grad_norm": 1.390625,
+      "learning_rate": 1.1802222222222223e-05,
+      "loss": 0.0384,
+      "mean_token_accuracy": 0.9892394132912159,
+      "num_tokens": 134198376.0,
+      "step": 4690
+    },
+    {
+      "entropy": 0.0634507178561762,
+      "epoch": 7.52,
+      "grad_norm": 0.7265625,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9894102096557618,
+      "num_tokens": 134491162.0,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_math_entropy": 0.19578898119926452,
+      "eval_math_loss": 1.2300976514816284,
+      "eval_math_mean_token_accuracy": 0.8157597880363464,
+      "eval_math_num_tokens": 134491162.0,
+      "eval_math_runtime": 35.7866,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4700
+    },
+    {
+      "epoch": 7.52,
+      "eval_physics_entropy": 0.21874468064308167,
+      "eval_physics_loss": 1.310891032218933,
+      "eval_physics_mean_token_accuracy": 0.7968743557929993,
+      "eval_physics_num_tokens": 134491162.0,
+      "eval_physics_runtime": 41.4806,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 4700
+    },
+    {
+      "entropy": 0.06613360948394983,
+      "epoch": 7.536,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.1757777777777778e-05,
+      "loss": 0.0397,
+      "mean_token_accuracy": 0.9887725580483675,
+      "num_tokens": 134772647.0,
+      "step": 4710
+    },
+    {
+      "entropy": 0.06250753803178669,
+      "epoch": 7.552,
+      "grad_norm": 0.6484375,
+      "learning_rate": 1.1735555555555556e-05,
+      "loss": 0.0381,
+      "mean_token_accuracy": 0.9894171375781298,
+      "num_tokens": 135062793.0,
+      "step": 4720
+    },
+    {
+      "entropy": 0.06458577974699438,
+      "epoch": 7.568,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.1713333333333334e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9891802627593279,
+      "num_tokens": 135352954.0,
+      "step": 4730
+    },
+    {
+      "entropy": 0.06382402395829559,
+      "epoch": 7.584,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.1691111111111113e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9893467653542757,
+      "num_tokens": 135643709.0,
+      "step": 4740
+    },
+    {
+      "entropy": 0.062360757333226505,
+      "epoch": 7.6,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.1668888888888889e-05,
+      "loss": 0.0382,
+      "mean_token_accuracy": 0.9892903961241245,
+      "num_tokens": 135933730.0,
+      "step": 4750
+    },
+    {
+      "entropy": 0.06553376447409391,
+      "epoch": 7.616,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.1646666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887435000389815,
+      "num_tokens": 136215545.0,
+      "step": 4760
+    },
+    {
+      "entropy": 0.06385163033846766,
+      "epoch": 7.632,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1624444444444446e-05,
+      "loss": 0.0379,
+      "mean_token_accuracy": 0.989441429823637,
+      "num_tokens": 136509697.0,
+      "step": 4770
+    },
+    {
+      "entropy": 0.06461194895673543,
+      "epoch": 7.648,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1602222222222223e-05,
+      "loss": 0.0398,
+      "mean_token_accuracy": 0.9889575779438019,
+      "num_tokens": 136794834.0,
+      "step": 4780
+    },
+    {
+      "entropy": 0.06364777330309153,
+      "epoch": 7.664,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9890321586281061,
+      "num_tokens": 137077698.0,
+      "step": 4790
+    },
+    {
+      "entropy": 0.06495753165800125,
+      "epoch": 7.68,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.1557777777777779e-05,
+      "loss": 0.0407,
+      "mean_token_accuracy": 0.9888042062520981,
+      "num_tokens": 137354100.0,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_math_entropy": 0.1961505770087242,
+      "eval_math_loss": 1.2280024290084839,
+      "eval_math_mean_token_accuracy": 0.8158598508834839,
+      "eval_math_num_tokens": 137354100.0,
+      "eval_math_runtime": 35.7851,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 4800
+    },
+    {
+      "epoch": 7.68,
+      "eval_physics_entropy": 0.21946947872638703,
+      "eval_physics_loss": 1.3097177743911743,
+      "eval_physics_mean_token_accuracy": 0.7968477377891541,
+      "eval_physics_num_tokens": 137354100.0,
+      "eval_physics_runtime": 41.4785,
+      "eval_physics_samples_per_second": 12.054,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 4800
+    },
+    {
+      "entropy": 0.06376796101685614,
+      "epoch": 7.696,
+      "grad_norm": 0.66015625,
+      "learning_rate": 1.1535555555555556e-05,
+      "loss": 0.0387,
+      "mean_token_accuracy": 0.9891933016479015,
+      "num_tokens": 137646688.0,
+      "step": 4810
+    },
+    {
+      "entropy": 0.06496104435063899,
+      "epoch": 7.712,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.1513333333333334e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888025749474764,
+      "num_tokens": 137930478.0,
+      "step": 4820
+    },
+    {
+      "entropy": 0.06421369009185582,
+      "epoch": 7.728,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1491111111111113e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9891324803233147,
+      "num_tokens": 138219528.0,
+      "step": 4830
+    },
+    {
+      "entropy": 0.06491015092469751,
+      "epoch": 7.744,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.1468888888888889e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.9887373331934214,
+      "num_tokens": 138505271.0,
+      "step": 4840
+    },
+    {
+      "entropy": 0.06559491397347302,
+      "epoch": 7.76,
+      "grad_norm": 0.76953125,
+      "learning_rate": 1.1446666666666668e-05,
+      "loss": 0.0396,
+      "mean_token_accuracy": 0.9887735281139612,
+      "num_tokens": 138790612.0,
+      "step": 4850
+    },
+    {
+      "entropy": 0.06484551811590791,
+      "epoch": 7.776,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1424444444444444e-05,
+      "loss": 0.0383,
+      "mean_token_accuracy": 0.9889496352523566,
+      "num_tokens": 139081356.0,
+      "step": 4860
+    },
+    {
+      "entropy": 0.06331797884777188,
+      "epoch": 7.792,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1402222222222224e-05,
+      "loss": 0.0372,
+      "mean_token_accuracy": 0.9895282920449973,
+      "num_tokens": 139380135.0,
+      "step": 4870
+    },
+    {
+      "entropy": 0.06256569258403033,
+      "epoch": 7.808,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.138e-05,
+      "loss": 0.0388,
+      "mean_token_accuracy": 0.9891048073768616,
+      "num_tokens": 139670791.0,
+      "step": 4880
+    },
+    {
+      "entropy": 0.06474397722631693,
+      "epoch": 7.824,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.1357777777777779e-05,
+      "loss": 0.0389,
+      "mean_token_accuracy": 0.988869558647275,
+      "num_tokens": 139954935.0,
+      "step": 4890
+    },
+    {
+      "entropy": 0.06261393134482204,
+      "epoch": 7.84,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1335555555555558e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9891028236597776,
+      "num_tokens": 140243499.0,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_math_entropy": 0.19476979553699494,
+      "eval_math_loss": 1.2364224195480347,
+      "eval_math_mean_token_accuracy": 0.8157722191810608,
+      "eval_math_num_tokens": 140243499.0,
+      "eval_math_runtime": 35.7988,
+      "eval_math_samples_per_second": 13.967,
+      "eval_math_steps_per_second": 3.492,
+      "step": 4900
+    },
+    {
+      "epoch": 7.84,
+      "eval_physics_entropy": 0.2179366738796234,
+      "eval_physics_loss": 1.3180210590362549,
+      "eval_physics_mean_token_accuracy": 0.7968727760314941,
+      "eval_physics_num_tokens": 140243499.0,
+      "eval_physics_runtime": 41.4652,
+      "eval_physics_samples_per_second": 12.058,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 4900
+    },
+    {
+      "entropy": 0.0660986915230751,
+      "epoch": 7.856,
+      "grad_norm": 0.796875,
+      "learning_rate": 1.1313333333333334e-05,
+      "loss": 0.0395,
+      "mean_token_accuracy": 0.9887882888317108,
+      "num_tokens": 140523048.0,
+      "step": 4910
+    },
+    {
+      "entropy": 0.06733772295992821,
+      "epoch": 7.872,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.1291111111111113e-05,
+      "loss": 0.0411,
+      "mean_token_accuracy": 0.9885444976389408,
+      "num_tokens": 140798987.0,
+      "step": 4920
+    },
+    {
+      "entropy": 0.06379716179799289,
+      "epoch": 7.888,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.126888888888889e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888154394924641,
+      "num_tokens": 141086821.0,
+      "step": 4930
+    },
+    {
+      "entropy": 0.06268583498895168,
+      "epoch": 7.904,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.1246666666666669e-05,
+      "loss": 0.0377,
+      "mean_token_accuracy": 0.9894758500158787,
+      "num_tokens": 141381172.0,
+      "step": 4940
+    },
+    {
+      "entropy": 0.06482646651566029,
+      "epoch": 7.92,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1224444444444444e-05,
+      "loss": 0.039,
+      "mean_token_accuracy": 0.9889786690473557,
+      "num_tokens": 141668933.0,
+      "step": 4950
+    },
+    {
+      "entropy": 0.06496006473898888,
+      "epoch": 7.936,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.1202222222222224e-05,
+      "loss": 0.0393,
+      "mean_token_accuracy": 0.9888488940894604,
+      "num_tokens": 141951435.0,
+      "step": 4960
+    },
+    {
+      "entropy": 0.06452734193298966,
+      "epoch": 7.952,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.1180000000000001e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9889313533902169,
+      "num_tokens": 142238865.0,
+      "step": 4970
+    },
+    {
+      "entropy": 0.06580421230755747,
+      "epoch": 7.968,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.1157777777777779e-05,
+      "loss": 0.04,
+      "mean_token_accuracy": 0.9887514386326075,
+      "num_tokens": 142519525.0,
+      "step": 4980
+    },
+    {
+      "entropy": 0.06502947660628706,
+      "epoch": 7.984,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.1135555555555557e-05,
+      "loss": 0.0391,
+      "mean_token_accuracy": 0.9888614844530821,
+      "num_tokens": 142805078.0,
+      "step": 4990
+    },
+    {
+      "entropy": 0.06425679824315012,
+      "epoch": 8.0,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.1113333333333334e-05,
+      "loss": 0.0394,
+      "mean_token_accuracy": 0.9888634238392114,
+      "num_tokens": 143091392.0,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_math_entropy": 0.1938511391878128,
+      "eval_math_loss": 1.2406725883483887,
+      "eval_math_mean_token_accuracy": 0.8157460641860962,
+      "eval_math_num_tokens": 143091392.0,
+      "eval_math_runtime": 35.8187,
+      "eval_math_samples_per_second": 13.959,
+      "eval_math_steps_per_second": 3.49,
+      "step": 5000
+    },
+    {
+      "epoch": 8.0,
+      "eval_physics_entropy": 0.21758921098709105,
+      "eval_physics_loss": 1.3192702531814575,
+      "eval_physics_mean_token_accuracy": 0.7965651154518127,
+      "eval_physics_num_tokens": 143091392.0,
+      "eval_physics_runtime": 41.2015,
+      "eval_physics_samples_per_second": 12.135,
+      "eval_physics_steps_per_second": 3.034,
+      "step": 5000
+    },
+    {
+      "entropy": 0.0571089421864599,
+      "epoch": 8.016,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.1091111111111112e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905061088502407,
+      "num_tokens": 143380616.0,
+      "step": 5010
+    },
+    {
+      "entropy": 0.056837924383580686,
+      "epoch": 8.032,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.106888888888889e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9903788235038519,
+      "num_tokens": 143664991.0,
+      "step": 5020
+    },
+    {
+      "entropy": 0.055355672526638956,
+      "epoch": 8.048,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.1046666666666667e-05,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9905342027544976,
+      "num_tokens": 143954009.0,
+      "step": 5030
+    },
+    {
+      "entropy": 0.05773321201559156,
+      "epoch": 8.064,
+      "grad_norm": 0.474609375,
+      "learning_rate": 1.1024444444444445e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898570664227009,
+      "num_tokens": 144234863.0,
+      "step": 5040
+    },
+    {
+      "entropy": 0.057087833108380434,
+      "epoch": 8.08,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.1002222222222222e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9902020487934351,
+      "num_tokens": 144520095.0,
+      "step": 5050
+    },
+    {
+      "entropy": 0.05786000648513436,
+      "epoch": 8.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.0980000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9902658749371767,
+      "num_tokens": 144807443.0,
+      "step": 5060
+    },
+    {
+      "entropy": 0.05588258268544451,
+      "epoch": 8.112,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.0957777777777778e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9904146701097488,
+      "num_tokens": 145096080.0,
+      "step": 5070
+    },
+    {
+      "entropy": 0.05747920935973525,
+      "epoch": 8.128,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.0935555555555557e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899541404098272,
+      "num_tokens": 145378174.0,
+      "step": 5080
+    },
+    {
+      "entropy": 0.05695802301634103,
+      "epoch": 8.144,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0913333333333333e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9901866424828768,
+      "num_tokens": 145663178.0,
+      "step": 5090
+    },
+    {
+      "entropy": 0.055338537762872876,
+      "epoch": 8.16,
+      "grad_norm": 0.37109375,
+      "learning_rate": 1.0891111111111112e-05,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.990602295473218,
+      "num_tokens": 145957628.0,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_math_entropy": 0.18742996579408647,
+      "eval_math_loss": 1.2740265130996704,
+      "eval_math_mean_token_accuracy": 0.8160112338066101,
+      "eval_math_num_tokens": 145957628.0,
+      "eval_math_runtime": 35.7701,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.495,
+      "step": 5100
+    },
+    {
+      "epoch": 8.16,
+      "eval_physics_entropy": 0.21262209969758988,
+      "eval_physics_loss": 1.3490732908248901,
+      "eval_physics_mean_token_accuracy": 0.7966427960395813,
+      "eval_physics_num_tokens": 145957628.0,
+      "eval_physics_runtime": 41.4455,
+      "eval_physics_samples_per_second": 12.064,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 5100
+    },
+    {
+      "entropy": 0.055332006863318385,
+      "epoch": 8.176,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0868888888888888e-05,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9903642062097788,
+      "num_tokens": 146247646.0,
+      "step": 5110
+    },
+    {
+      "entropy": 0.056531890854239465,
+      "epoch": 8.192,
+      "grad_norm": 0.5703125,
+      "learning_rate": 1.0846666666666667e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902558472007513,
+      "num_tokens": 146534877.0,
+      "step": 5120
+    },
+    {
+      "entropy": 0.05871388749219477,
+      "epoch": 8.208,
+      "grad_norm": 0.53125,
+      "learning_rate": 1.0824444444444447e-05,
+      "loss": 0.0344,
+      "mean_token_accuracy": 0.9898561745882034,
+      "num_tokens": 146814001.0,
+      "step": 5130
+    },
+    {
+      "entropy": 0.05695201009511948,
+      "epoch": 8.224,
+      "grad_norm": 0.60546875,
+      "learning_rate": 1.0802222222222223e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9899209447205066,
+      "num_tokens": 147095704.0,
+      "step": 5140
+    },
+    {
+      "entropy": 0.05595462932251394,
+      "epoch": 8.24,
+      "grad_norm": 0.451171875,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9900648534297943,
+      "num_tokens": 147385308.0,
+      "step": 5150
+    },
+    {
+      "entropy": 0.05648714422713965,
+      "epoch": 8.256,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.0757777777777778e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900227766484022,
+      "num_tokens": 147675332.0,
+      "step": 5160
+    },
+    {
+      "entropy": 0.057042692345567046,
+      "epoch": 8.272,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.0735555555555557e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900161389261484,
+      "num_tokens": 147963090.0,
+      "step": 5170
+    },
+    {
+      "entropy": 0.057354798656888306,
+      "epoch": 8.288,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0713333333333333e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898492351174355,
+      "num_tokens": 148246309.0,
+      "step": 5180
+    },
+    {
+      "entropy": 0.056936953822150825,
+      "epoch": 8.304,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.0691111111111112e-05,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9899033553898334,
+      "num_tokens": 148532110.0,
+      "step": 5190
+    },
+    {
+      "entropy": 0.05525723719038069,
+      "epoch": 8.32,
+      "grad_norm": 0.423828125,
+      "learning_rate": 1.0668888888888892e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9903490521013737,
+      "num_tokens": 148821239.0,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_math_entropy": 0.18771023988723756,
+      "eval_math_loss": 1.2761003971099854,
+      "eval_math_mean_token_accuracy": 0.8161793885231018,
+      "eval_math_num_tokens": 148821239.0,
+      "eval_math_runtime": 35.7715,
+      "eval_math_samples_per_second": 13.978,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5200
+    },
+    {
+      "epoch": 8.32,
+      "eval_physics_entropy": 0.2129958848953247,
+      "eval_physics_loss": 1.3503196239471436,
+      "eval_physics_mean_token_accuracy": 0.7967255554199218,
+      "eval_physics_num_tokens": 148821239.0,
+      "eval_physics_runtime": 41.4715,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 5200
+    },
+    {
+      "entropy": 0.05677824043668807,
+      "epoch": 8.336,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.0646666666666668e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899443361908198,
+      "num_tokens": 149104733.0,
+      "step": 5210
+    },
+    {
+      "entropy": 0.05626459107734263,
+      "epoch": 8.352,
+      "grad_norm": 0.46875,
+      "learning_rate": 1.0624444444444447e-05,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9899135671555996,
+      "num_tokens": 149392779.0,
+      "step": 5220
+    },
+    {
+      "entropy": 0.056180683872662486,
+      "epoch": 8.368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.0602222222222223e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.990152184665203,
+      "num_tokens": 149680564.0,
+      "step": 5230
+    },
+    {
+      "entropy": 0.05665554238948971,
+      "epoch": 8.384,
+      "grad_norm": 0.67578125,
+      "learning_rate": 1.0580000000000002e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9902399625629187,
+      "num_tokens": 149974415.0,
+      "step": 5240
+    },
+    {
+      "entropy": 0.05632798650767654,
+      "epoch": 8.4,
+      "grad_norm": 0.48046875,
+      "learning_rate": 1.0557777777777778e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900553777813912,
+      "num_tokens": 150263671.0,
+      "step": 5250
+    },
+    {
+      "entropy": 0.05675367978401482,
+      "epoch": 8.416,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0535555555555557e-05,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.990035155043006,
+      "num_tokens": 150554357.0,
+      "step": 5260
+    },
+    {
+      "entropy": 0.05795961414696649,
+      "epoch": 8.432,
+      "grad_norm": 0.54296875,
+      "learning_rate": 1.0513333333333333e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896357730031013,
+      "num_tokens": 150832925.0,
+      "step": 5270
+    },
+    {
+      "entropy": 0.05585661982186139,
+      "epoch": 8.448,
+      "grad_norm": 0.5625,
+      "learning_rate": 1.0491111111111112e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9902568813413382,
+      "num_tokens": 151118562.0,
+      "step": 5280
+    },
+    {
+      "entropy": 0.0565168714732863,
+      "epoch": 8.464,
+      "grad_norm": 0.47265625,
+      "learning_rate": 1.046888888888889e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898991256952285,
+      "num_tokens": 151407229.0,
+      "step": 5290
+    },
+    {
+      "entropy": 0.055777230369858447,
+      "epoch": 8.48,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0446666666666668e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9900368627160787,
+      "num_tokens": 151695842.0,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_math_entropy": 0.1862943674325943,
+      "eval_math_loss": 1.2794464826583862,
+      "eval_math_mean_token_accuracy": 0.8162452592849732,
+      "eval_math_num_tokens": 151695842.0,
+      "eval_math_runtime": 35.8249,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5300
+    },
+    {
+      "epoch": 8.48,
+      "eval_physics_entropy": 0.211733931183815,
+      "eval_physics_loss": 1.3531311750411987,
+      "eval_physics_mean_token_accuracy": 0.7966176652908326,
+      "eval_physics_num_tokens": 151695842.0,
+      "eval_physics_runtime": 41.509,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5300
+    },
+    {
+      "entropy": 0.05595301657449454,
+      "epoch": 8.496,
+      "grad_norm": 0.490234375,
+      "learning_rate": 1.0424444444444445e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900278240442276,
+      "num_tokens": 151981315.0,
+      "step": 5310
+    },
+    {
+      "entropy": 0.056305208592675626,
+      "epoch": 8.512,
+      "grad_norm": 0.49609375,
+      "learning_rate": 1.0402222222222223e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9898664347827435,
+      "num_tokens": 152268106.0,
+      "step": 5320
+    },
+    {
+      "entropy": 0.05696632086765021,
+      "epoch": 8.528,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9899542909115553,
+      "num_tokens": 152553174.0,
+      "step": 5330
+    },
+    {
+      "entropy": 0.05757405322510749,
+      "epoch": 8.544,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.0357777777777778e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9899792809039354,
+      "num_tokens": 152836031.0,
+      "step": 5340
+    },
+    {
+      "entropy": 0.05707665903028101,
+      "epoch": 8.56,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.0335555555555556e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9898822121322155,
+      "num_tokens": 153118518.0,
+      "step": 5350
+    },
+    {
+      "entropy": 0.05585220886860043,
+      "epoch": 8.576,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0313333333333335e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9901655036956072,
+      "num_tokens": 153406255.0,
+      "step": 5360
+    },
+    {
+      "entropy": 0.05792757070157677,
+      "epoch": 8.592,
+      "grad_norm": 0.703125,
+      "learning_rate": 1.0291111111111111e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897722654044628,
+      "num_tokens": 153691609.0,
+      "step": 5370
+    },
+    {
+      "entropy": 0.05656812160741538,
+      "epoch": 8.608,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.026888888888889e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9899586401879787,
+      "num_tokens": 153972327.0,
+      "step": 5380
+    },
+    {
+      "entropy": 0.05814382419921458,
+      "epoch": 8.624,
+      "grad_norm": 0.56640625,
+      "learning_rate": 1.0246666666666666e-05,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9896783825010061,
+      "num_tokens": 154253859.0,
+      "step": 5390
+    },
+    {
+      "entropy": 0.056846255506388846,
+      "epoch": 8.64,
+      "grad_norm": 0.5390625,
+      "learning_rate": 1.0224444444444446e-05,
+      "loss": 0.0351,
+      "mean_token_accuracy": 0.9897733077406883,
+      "num_tokens": 154533424.0,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_math_entropy": 0.18660490876436234,
+      "eval_math_loss": 1.281345009803772,
+      "eval_math_mean_token_accuracy": 0.8162109370231628,
+      "eval_math_num_tokens": 154533424.0,
+      "eval_math_runtime": 35.773,
+      "eval_math_samples_per_second": 13.977,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5400
+    },
+    {
+      "epoch": 8.64,
+      "eval_physics_entropy": 0.21224513775110246,
+      "eval_physics_loss": 1.3543466329574585,
+      "eval_physics_mean_token_accuracy": 0.7967072949409485,
+      "eval_physics_num_tokens": 154533424.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 5400
+    },
+    {
+      "entropy": 0.05667855739593506,
+      "epoch": 8.656,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.0202222222222221e-05,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9901769265532494,
+      "num_tokens": 154821917.0,
+      "step": 5410
+    },
+    {
+      "entropy": 0.05516336506698281,
+      "epoch": 8.672,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9904799718409777,
+      "num_tokens": 155116902.0,
+      "step": 5420
+    },
+    {
+      "entropy": 0.05663891236763448,
+      "epoch": 8.688,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.0157777777777777e-05,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899812191724777,
+      "num_tokens": 155407148.0,
+      "step": 5430
+    },
+    {
+      "entropy": 0.05660432383883744,
+      "epoch": 8.704,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.0135555555555556e-05,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.9898883264511824,
+      "num_tokens": 155693453.0,
+      "step": 5440
+    },
+    {
+      "entropy": 0.05770897620823234,
+      "epoch": 8.72,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.0113333333333335e-05,
+      "loss": 0.035,
+      "mean_token_accuracy": 0.9894243434071541,
+      "num_tokens": 155973181.0,
+      "step": 5450
+    },
+    {
+      "entropy": 0.05790396251250059,
+      "epoch": 8.736,
+      "grad_norm": 0.6015625,
+      "learning_rate": 1.0091111111111111e-05,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.9896433904767037,
+      "num_tokens": 156253770.0,
+      "step": 5460
+    },
+    {
+      "entropy": 0.056275642523542044,
+      "epoch": 8.752,
+      "grad_norm": 0.578125,
+      "learning_rate": 1.006888888888889e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9899505387991667,
+      "num_tokens": 156539835.0,
+      "step": 5470
+    },
+    {
+      "entropy": 0.056800450780428946,
+      "epoch": 8.768,
+      "grad_norm": 0.453125,
+      "learning_rate": 1.0046666666666666e-05,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898199662566185,
+      "num_tokens": 156827084.0,
+      "step": 5480
+    },
+    {
+      "entropy": 0.05637329164892435,
+      "epoch": 8.784,
+      "grad_norm": 0.44140625,
+      "learning_rate": 1.0024444444444446e-05,
+      "loss": 0.034,
+      "mean_token_accuracy": 0.9899186108261346,
+      "num_tokens": 157116182.0,
+      "step": 5490
+    },
+    {
+      "entropy": 0.05638529262505472,
+      "epoch": 8.8,
+      "grad_norm": 0.6328125,
+      "learning_rate": 1.0002222222222222e-05,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9898790910840034,
+      "num_tokens": 157403738.0,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_math_entropy": 0.18709938406944274,
+      "eval_math_loss": 1.282743215560913,
+      "eval_math_mean_token_accuracy": 0.8162371826171875,
+      "eval_math_num_tokens": 157403738.0,
+      "eval_math_runtime": 35.7777,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 5500
+    },
+    {
+      "epoch": 8.8,
+      "eval_physics_entropy": 0.21249422907829285,
+      "eval_physics_loss": 1.3553577661514282,
+      "eval_physics_mean_token_accuracy": 0.7966284174919128,
+      "eval_physics_num_tokens": 157403738.0,
+      "eval_physics_runtime": 41.1721,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 5500
+    },
+    {
+      "entropy": 0.057882068841718134,
+      "epoch": 8.816,
+      "grad_norm": 0.5234375,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 0.0352,
+      "mean_token_accuracy": 0.989596389606595,
+      "num_tokens": 157682311.0,
+      "step": 5510
+    },
+    {
+      "entropy": 0.05714819370768964,
+      "epoch": 8.832,
+      "grad_norm": 0.3828125,
+      "learning_rate": 9.957777777777779e-06,
+      "loss": 0.0347,
+      "mean_token_accuracy": 0.9897560972720385,
+      "num_tokens": 157969448.0,
+      "step": 5520
+    },
+    {
+      "entropy": 0.057319192099384964,
+      "epoch": 8.848,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.935555555555556e-06,
+      "loss": 0.0353,
+      "mean_token_accuracy": 0.9893435817211866,
+      "num_tokens": 158253721.0,
+      "step": 5530
+    },
+    {
+      "entropy": 0.05722528274636716,
+      "epoch": 8.864,
+      "grad_norm": 0.51171875,
+      "learning_rate": 9.913333333333334e-06,
+      "loss": 0.0346,
+      "mean_token_accuracy": 0.9897772017866373,
+      "num_tokens": 158538794.0,
+      "step": 5540
+    },
+    {
+      "entropy": 0.05734995815437287,
+      "epoch": 8.88,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.891111111111113e-06,
+      "loss": 0.0349,
+      "mean_token_accuracy": 0.9897694259881973,
+      "num_tokens": 158823829.0,
+      "step": 5550
+    },
+    {
+      "entropy": 0.057489245175383986,
+      "epoch": 8.896,
+      "grad_norm": 0.578125,
+      "learning_rate": 9.86888888888889e-06,
+      "loss": 0.0348,
+      "mean_token_accuracy": 0.9898175619542599,
+      "num_tokens": 159106769.0,
+      "step": 5560
+    },
+    {
+      "entropy": 0.054861981072463095,
+      "epoch": 8.912,
+      "grad_norm": 0.5078125,
+      "learning_rate": 9.846666666666668e-06,
+      "loss": 0.0341,
+      "mean_token_accuracy": 0.9900507286190987,
+      "num_tokens": 159395580.0,
+      "step": 5570
+    },
+    {
+      "entropy": 0.056686159083619715,
+      "epoch": 8.928,
+      "grad_norm": 0.494140625,
+      "learning_rate": 9.824444444444446e-06,
+      "loss": 0.0342,
+      "mean_token_accuracy": 0.9900229871273041,
+      "num_tokens": 159685319.0,
+      "step": 5580
+    },
+    {
+      "entropy": 0.05702824930194765,
+      "epoch": 8.943999999999999,
+      "grad_norm": 0.47265625,
+      "learning_rate": 9.802222222222224e-06,
+      "loss": 0.0343,
+      "mean_token_accuracy": 0.9898402344435453,
+      "num_tokens": 159972270.0,
+      "step": 5590
+    },
+    {
+      "entropy": 0.05834557798225433,
+      "epoch": 8.96,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 0.0358,
+      "mean_token_accuracy": 0.9893978051841259,
+      "num_tokens": 160249370.0,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_math_entropy": 0.18625934714078904,
+      "eval_math_loss": 1.284448504447937,
+      "eval_math_mean_token_accuracy": 0.816104745388031,
+      "eval_math_num_tokens": 160249370.0,
+      "eval_math_runtime": 35.8291,
+      "eval_math_samples_per_second": 13.955,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5600
+    },
+    {
+      "epoch": 8.96,
+      "eval_physics_entropy": 0.2115949323773384,
+      "eval_physics_loss": 1.3571308851242065,
+      "eval_physics_mean_token_accuracy": 0.796782205581665,
+      "eval_physics_num_tokens": 160249370.0,
+      "eval_physics_runtime": 41.5164,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 5600
+    },
+    {
+      "entropy": 0.05641181897372007,
+      "epoch": 8.975999999999999,
+      "grad_norm": 0.498046875,
+      "learning_rate": 9.757777777777779e-06,
+      "loss": 0.0345,
+      "mean_token_accuracy": 0.990053978189826,
+      "num_tokens": 160535576.0,
+      "step": 5610
+    },
+    {
+      "entropy": 0.054938315274193884,
+      "epoch": 8.992,
+      "grad_norm": 0.546875,
+      "learning_rate": 9.735555555555556e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9904120992869139,
+      "num_tokens": 160826704.0,
+      "step": 5620
+    },
+    {
+      "entropy": 0.05459052384831011,
+      "epoch": 9.008,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.713333333333334e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9905484411865473,
+      "num_tokens": 161124282.0,
+      "step": 5630
+    },
+    {
+      "entropy": 0.056366299442015586,
+      "epoch": 9.024,
+      "grad_norm": 0.439453125,
+      "learning_rate": 9.691111111111112e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.99030572026968,
+      "num_tokens": 161403247.0,
+      "step": 5640
+    },
+    {
+      "entropy": 0.05284358374774456,
+      "epoch": 9.04,
+      "grad_norm": 0.455078125,
+      "learning_rate": 9.66888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990729172155261,
+      "num_tokens": 161687815.0,
+      "step": 5650
+    },
+    {
+      "entropy": 0.05345777831971645,
+      "epoch": 9.056,
+      "grad_norm": 0.45703125,
+      "learning_rate": 9.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990471850708127,
+      "num_tokens": 161971275.0,
+      "step": 5660
+    },
+    {
+      "entropy": 0.053053504962008446,
+      "epoch": 9.072,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.624444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907864902168513,
+      "num_tokens": 162264276.0,
+      "step": 5670
+    },
+    {
+      "entropy": 0.05359348710626364,
+      "epoch": 9.088,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.602222222222222e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904586333781481,
+      "num_tokens": 162551120.0,
+      "step": 5680
+    },
+    {
+      "entropy": 0.05278733624145389,
+      "epoch": 9.104,
+      "grad_norm": 0.40625,
+      "learning_rate": 9.58e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990637419745326,
+      "num_tokens": 162839051.0,
+      "step": 5690
+    },
+    {
+      "entropy": 0.055156051134690645,
+      "epoch": 9.12,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.557777777777777e-06,
+      "loss": 0.0333,
+      "mean_token_accuracy": 0.9902046132832766,
+      "num_tokens": 163114901.0,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_math_entropy": 0.18412417340278625,
+      "eval_math_loss": 1.3002654314041138,
+      "eval_math_mean_token_accuracy": 0.8162156238555908,
+      "eval_math_num_tokens": 163114901.0,
+      "eval_math_runtime": 35.8246,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 5700
+    },
+    {
+      "epoch": 9.12,
+      "eval_physics_entropy": 0.2103169944882393,
+      "eval_physics_loss": 1.3711953163146973,
+      "eval_physics_mean_token_accuracy": 0.7965371232032776,
+      "eval_physics_num_tokens": 163114901.0,
+      "eval_physics_runtime": 41.5576,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 5700
+    },
+    {
+      "entropy": 0.05398498624563217,
+      "epoch": 9.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 9.535555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906857419759035,
+      "num_tokens": 163408368.0,
+      "step": 5710
+    },
+    {
+      "entropy": 0.05103533309884369,
+      "epoch": 9.152,
+      "grad_norm": 0.380859375,
+      "learning_rate": 9.513333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9911101579666137,
+      "num_tokens": 163709408.0,
+      "step": 5720
+    },
+    {
+      "entropy": 0.05408018168527633,
+      "epoch": 9.168,
+      "grad_norm": 0.408203125,
+      "learning_rate": 9.491111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903054315596819,
+      "num_tokens": 163993409.0,
+      "step": 5730
+    },
+    {
+      "entropy": 0.05453936075791717,
+      "epoch": 9.184,
+      "grad_norm": 0.392578125,
+      "learning_rate": 9.46888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903354767709971,
+      "num_tokens": 164277822.0,
+      "step": 5740
+    },
+    {
+      "entropy": 0.05500368729699403,
+      "epoch": 9.2,
+      "grad_norm": 0.474609375,
+      "learning_rate": 9.446666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901658721268177,
+      "num_tokens": 164562030.0,
+      "step": 5750
+    },
+    {
+      "entropy": 0.05307849880773574,
+      "epoch": 9.216,
+      "grad_norm": 0.5625,
+      "learning_rate": 9.424444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906642295420169,
+      "num_tokens": 164857220.0,
+      "step": 5760
+    },
+    {
+      "entropy": 0.0552012657513842,
+      "epoch": 9.232,
+      "grad_norm": 0.462890625,
+      "learning_rate": 9.402222222222222e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9901241518557071,
+      "num_tokens": 165134875.0,
+      "step": 5770
+    },
+    {
+      "entropy": 0.0541033823043108,
+      "epoch": 9.248,
+      "grad_norm": 0.62890625,
+      "learning_rate": 9.38e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902011703699827,
+      "num_tokens": 165419679.0,
+      "step": 5780
+    },
+    {
+      "entropy": 0.05229763046372682,
+      "epoch": 9.264,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.35777777777778e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906760297715664,
+      "num_tokens": 165711282.0,
+      "step": 5790
+    },
+    {
+      "entropy": 0.05538548887707293,
+      "epoch": 9.28,
+      "grad_norm": 0.43359375,
+      "learning_rate": 9.335555555555557e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9900773245841265,
+      "num_tokens": 165988534.0,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_math_entropy": 0.18346527618169783,
+      "eval_math_loss": 1.3040590286254883,
+      "eval_math_mean_token_accuracy": 0.8161579728126526,
+      "eval_math_num_tokens": 165988534.0,
+      "eval_math_runtime": 35.854,
+      "eval_math_samples_per_second": 13.945,
+      "eval_math_steps_per_second": 3.486,
+      "step": 5800
+    },
+    {
+      "epoch": 9.28,
+      "eval_physics_entropy": 0.20937572026252746,
+      "eval_physics_loss": 1.3746147155761719,
+      "eval_physics_mean_token_accuracy": 0.7967208256721496,
+      "eval_physics_num_tokens": 165988534.0,
+      "eval_physics_runtime": 41.5441,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5800
+    },
+    {
+      "entropy": 0.05415816385066137,
+      "epoch": 9.296,
+      "grad_norm": 0.443359375,
+      "learning_rate": 9.313333333333335e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9903882455080748,
+      "num_tokens": 166274754.0,
+      "step": 5810
+    },
+    {
+      "entropy": 0.05424636383540928,
+      "epoch": 9.312,
+      "grad_norm": 0.3984375,
+      "learning_rate": 9.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904342375695705,
+      "num_tokens": 166563722.0,
+      "step": 5820
+    },
+    {
+      "entropy": 0.053859639749862256,
+      "epoch": 9.328,
+      "grad_norm": 0.42578125,
+      "learning_rate": 9.26888888888889e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9902767721563578,
+      "num_tokens": 166848787.0,
+      "step": 5830
+    },
+    {
+      "entropy": 0.053050220373552295,
+      "epoch": 9.344,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.246666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9905906770378351,
+      "num_tokens": 167137520.0,
+      "step": 5840
+    },
+    {
+      "entropy": 0.05505580292083323,
+      "epoch": 9.36,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.224444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9899543590843678,
+      "num_tokens": 167424017.0,
+      "step": 5850
+    },
+    {
+      "entropy": 0.05398182098288089,
+      "epoch": 9.376,
+      "grad_norm": 0.44140625,
+      "learning_rate": 9.202222222222224e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903419081121683,
+      "num_tokens": 167710420.0,
+      "step": 5860
+    },
+    {
+      "entropy": 0.05449502062983811,
+      "epoch": 9.392,
+      "grad_norm": 0.46875,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9901338364928961,
+      "num_tokens": 167991686.0,
+      "step": 5870
+    },
+    {
+      "entropy": 0.05327932327054441,
+      "epoch": 9.408,
+      "grad_norm": 0.412109375,
+      "learning_rate": 9.15777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903248358517885,
+      "num_tokens": 168279941.0,
+      "step": 5880
+    },
+    {
+      "entropy": 0.05446169211063534,
+      "epoch": 9.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 9.135555555555557e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903256863355636,
+      "num_tokens": 168569139.0,
+      "step": 5890
+    },
+    {
+      "entropy": 0.05290604527108371,
+      "epoch": 9.44,
+      "grad_norm": 0.384765625,
+      "learning_rate": 9.113333333333335e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904143486171961,
+      "num_tokens": 168858673.0,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_math_entropy": 0.1840248066186905,
+      "eval_math_loss": 1.3032736778259277,
+      "eval_math_mean_token_accuracy": 0.8160810256004334,
+      "eval_math_num_tokens": 168858673.0,
+      "eval_math_runtime": 35.8367,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 5900
+    },
+    {
+      "epoch": 9.44,
+      "eval_physics_entropy": 0.2098150200843811,
+      "eval_physics_loss": 1.3739335536956787,
+      "eval_physics_mean_token_accuracy": 0.796475242137909,
+      "eval_physics_num_tokens": 168858673.0,
+      "eval_physics_runtime": 41.5473,
+      "eval_physics_samples_per_second": 12.034,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 5900
+    },
+    {
+      "entropy": 0.054311024211347106,
+      "epoch": 9.456,
+      "grad_norm": 0.349609375,
+      "learning_rate": 9.091111111111112e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903390396386385,
+      "num_tokens": 169146097.0,
+      "step": 5910
+    },
+    {
+      "entropy": 0.05390293239615858,
+      "epoch": 9.472,
+      "grad_norm": 0.421875,
+      "learning_rate": 9.06888888888889e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902336321771145,
+      "num_tokens": 169432094.0,
+      "step": 5920
+    },
+    {
+      "entropy": 0.0531583295436576,
+      "epoch": 9.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 9.046666666666668e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9906218215823174,
+      "num_tokens": 169728455.0,
+      "step": 5930
+    },
+    {
+      "entropy": 0.054046934354119,
+      "epoch": 9.504,
+      "grad_norm": 0.404296875,
+      "learning_rate": 9.024444444444445e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901666518300771,
+      "num_tokens": 170013591.0,
+      "step": 5940
+    },
+    {
+      "entropy": 0.05549235251965001,
+      "epoch": 9.52,
+      "grad_norm": 0.453125,
+      "learning_rate": 9.002222222222223e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9900341626256705,
+      "num_tokens": 170295820.0,
+      "step": 5950
+    },
+    {
+      "entropy": 0.05413213314022869,
+      "epoch": 9.536,
+      "grad_norm": 0.439453125,
+      "learning_rate": 8.98e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9902442399412393,
+      "num_tokens": 170582177.0,
+      "step": 5960
+    },
+    {
+      "entropy": 0.0534646904678084,
+      "epoch": 9.552,
+      "grad_norm": 0.41015625,
+      "learning_rate": 8.957777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901266142725944,
+      "num_tokens": 170869699.0,
+      "step": 5970
+    },
+    {
+      "entropy": 0.05319979584310204,
+      "epoch": 9.568,
+      "grad_norm": 0.427734375,
+      "learning_rate": 8.935555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9904231581836939,
+      "num_tokens": 171161292.0,
+      "step": 5980
+    },
+    {
+      "entropy": 0.055327297281473876,
+      "epoch": 9.584,
+      "grad_norm": 0.453125,
+      "learning_rate": 8.913333333333333e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990069554746151,
+      "num_tokens": 171445012.0,
+      "step": 5990
+    },
+    {
+      "entropy": 0.05277198071125895,
+      "epoch": 9.6,
+      "grad_norm": 0.38671875,
+      "learning_rate": 8.891111111111111e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904008969664574,
+      "num_tokens": 171740173.0,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_math_entropy": 0.1828007377386093,
+      "eval_math_loss": 1.3068181276321411,
+      "eval_math_mean_token_accuracy": 0.8161692705154419,
+      "eval_math_num_tokens": 171740173.0,
+      "eval_math_runtime": 35.8412,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 6000
+    },
+    {
+      "epoch": 9.6,
+      "eval_physics_entropy": 0.20892624753713607,
+      "eval_physics_loss": 1.375573754310608,
+      "eval_physics_mean_token_accuracy": 0.7965809445381165,
+      "eval_physics_num_tokens": 171740173.0,
+      "eval_physics_runtime": 41.1678,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 6000
+    },
+    {
+      "entropy": 0.05559574081562459,
+      "epoch": 9.616,
+      "grad_norm": 0.43359375,
+      "learning_rate": 8.868888888888888e-06,
+      "loss": 0.0338,
+      "mean_token_accuracy": 0.9898189619183541,
+      "num_tokens": 172017264.0,
+      "step": 6010
+    },
+    {
+      "entropy": 0.05323043402750045,
+      "epoch": 9.632,
+      "grad_norm": 0.5234375,
+      "learning_rate": 8.846666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.990172890946269,
+      "num_tokens": 172308431.0,
+      "step": 6020
+    },
+    {
+      "entropy": 0.055105213075876236,
+      "epoch": 9.648,
+      "grad_norm": 0.388671875,
+      "learning_rate": 8.824444444444445e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9898421704769135,
+      "num_tokens": 172589506.0,
+      "step": 6030
+    },
+    {
+      "entropy": 0.05343689962755889,
+      "epoch": 9.664,
+      "grad_norm": 0.390625,
+      "learning_rate": 8.802222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990496426448226,
+      "num_tokens": 172881834.0,
+      "step": 6040
+    },
+    {
+      "entropy": 0.05498569174669683,
+      "epoch": 9.68,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.78e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9900029398500919,
+      "num_tokens": 173163955.0,
+      "step": 6050
+    },
+    {
+      "entropy": 0.056193693052046,
+      "epoch": 9.696,
+      "grad_norm": 0.404296875,
+      "learning_rate": 8.757777777777778e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9899830743670464,
+      "num_tokens": 173440726.0,
+      "step": 6060
+    },
+    {
+      "entropy": 0.05423595421016216,
+      "epoch": 9.712,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.735555555555556e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901481702923774,
+      "num_tokens": 173726700.0,
+      "step": 6070
+    },
+    {
+      "entropy": 0.052887511567678305,
+      "epoch": 9.728,
+      "grad_norm": 0.3984375,
+      "learning_rate": 8.713333333333333e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903892632573843,
+      "num_tokens": 174021017.0,
+      "step": 6080
+    },
+    {
+      "entropy": 0.05478028848301619,
+      "epoch": 9.744,
+      "grad_norm": 0.423828125,
+      "learning_rate": 8.691111111111111e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899331051856279,
+      "num_tokens": 174303795.0,
+      "step": 6090
+    },
+    {
+      "entropy": 0.053893138212151824,
+      "epoch": 9.76,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900019478052855,
+      "num_tokens": 174593168.0,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_math_entropy": 0.18270208221673964,
+      "eval_math_loss": 1.3073022365570068,
+      "eval_math_mean_token_accuracy": 0.8161264581680298,
+      "eval_math_num_tokens": 174593168.0,
+      "eval_math_runtime": 35.7933,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6100
+    },
+    {
+      "epoch": 9.76,
+      "eval_physics_entropy": 0.20895446288585662,
+      "eval_physics_loss": 1.3762085437774658,
+      "eval_physics_mean_token_accuracy": 0.7966276407241821,
+      "eval_physics_num_tokens": 174593168.0,
+      "eval_physics_runtime": 41.4732,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 6100
+    },
+    {
+      "entropy": 0.05330368627328426,
+      "epoch": 9.776,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.646666666666668e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9900734055787325,
+      "num_tokens": 174875635.0,
+      "step": 6110
+    },
+    {
+      "entropy": 0.05547493495978415,
+      "epoch": 9.792,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.624444444444446e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.9899276282638312,
+      "num_tokens": 175160009.0,
+      "step": 6120
+    },
+    {
+      "entropy": 0.055491062253713606,
+      "epoch": 9.808,
+      "grad_norm": 0.443359375,
+      "learning_rate": 8.602222222222223e-06,
+      "loss": 0.0336,
+      "mean_token_accuracy": 0.9898974467068911,
+      "num_tokens": 175440699.0,
+      "step": 6130
+    },
+    {
+      "entropy": 0.05588666221592575,
+      "epoch": 9.824,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 0.0339,
+      "mean_token_accuracy": 0.9898178771138191,
+      "num_tokens": 175718579.0,
+      "step": 6140
+    },
+    {
+      "entropy": 0.05419871902558952,
+      "epoch": 9.84,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.557777777777778e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9901195026934146,
+      "num_tokens": 176004543.0,
+      "step": 6150
+    },
+    {
+      "entropy": 0.05438464351464063,
+      "epoch": 9.856,
+      "grad_norm": 0.44140625,
+      "learning_rate": 8.535555555555556e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.990145793929696,
+      "num_tokens": 176290637.0,
+      "step": 6160
+    },
+    {
+      "entropy": 0.054657111316919325,
+      "epoch": 9.872,
+      "grad_norm": 0.4140625,
+      "learning_rate": 8.513333333333335e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901409171521663,
+      "num_tokens": 176577140.0,
+      "step": 6170
+    },
+    {
+      "entropy": 0.05510451651643962,
+      "epoch": 9.888,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.491111111111113e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899644508957863,
+      "num_tokens": 176860079.0,
+      "step": 6180
+    },
+    {
+      "entropy": 0.05548927634954452,
+      "epoch": 9.904,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.46888888888889e-06,
+      "loss": 0.0337,
+      "mean_token_accuracy": 0.9899002809077502,
+      "num_tokens": 177140677.0,
+      "step": 6190
+    },
+    {
+      "entropy": 0.05508114048279822,
+      "epoch": 9.92,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.446666666666668e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9899649370461703,
+      "num_tokens": 177428933.0,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_math_entropy": 0.18255306565761567,
+      "eval_math_loss": 1.3089299201965332,
+      "eval_math_mean_token_accuracy": 0.8162662987709045,
+      "eval_math_num_tokens": 177428933.0,
+      "eval_math_runtime": 35.7967,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6200
+    },
+    {
+      "epoch": 9.92,
+      "eval_physics_entropy": 0.208550783097744,
+      "eval_physics_loss": 1.3781224489212036,
+      "eval_physics_mean_token_accuracy": 0.7966379327774048,
+      "eval_physics_num_tokens": 177428933.0,
+      "eval_physics_runtime": 41.4891,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6200
+    },
+    {
+      "entropy": 0.054707426903769375,
+      "epoch": 9.936,
+      "grad_norm": 0.375,
+      "learning_rate": 8.424444444444446e-06,
+      "loss": 0.0334,
+      "mean_token_accuracy": 0.9899840228259563,
+      "num_tokens": 177712917.0,
+      "step": 6210
+    },
+    {
+      "entropy": 0.053940073936246336,
+      "epoch": 9.952,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.402222222222223e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9901683274656534,
+      "num_tokens": 177999030.0,
+      "step": 6220
+    },
+    {
+      "entropy": 0.05412060760427266,
+      "epoch": 9.968,
+      "grad_norm": 0.4296875,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901083376258611,
+      "num_tokens": 178285808.0,
+      "step": 6230
+    },
+    {
+      "entropy": 0.054595653573051094,
+      "epoch": 9.984,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.357777777777779e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9899616222828627,
+      "num_tokens": 178575558.0,
+      "step": 6240
+    },
+    {
+      "entropy": 0.05434199175797403,
+      "epoch": 10.0,
+      "grad_norm": 0.384765625,
+      "learning_rate": 8.335555555555556e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9900276735424995,
+      "num_tokens": 178864240.0,
+      "step": 6250
+    },
+    {
+      "entropy": 0.05455364771187306,
+      "epoch": 10.016,
+      "grad_norm": 0.40234375,
+      "learning_rate": 8.313333333333334e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904673513025045,
+      "num_tokens": 179147535.0,
+      "step": 6260
+    },
+    {
+      "entropy": 0.05369720469461754,
+      "epoch": 10.032,
+      "grad_norm": 0.412109375,
+      "learning_rate": 8.291111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903138358145952,
+      "num_tokens": 179427545.0,
+      "step": 6270
+    },
+    {
+      "entropy": 0.05132723795250058,
+      "epoch": 10.048,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907003980129957,
+      "num_tokens": 179718602.0,
+      "step": 6280
+    },
+    {
+      "entropy": 0.051855225989129394,
+      "epoch": 10.064,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.246666666666667e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907560586929322,
+      "num_tokens": 180006589.0,
+      "step": 6290
+    },
+    {
+      "entropy": 0.05200373688712716,
+      "epoch": 10.08,
+      "grad_norm": 0.40625,
+      "learning_rate": 8.224444444444444e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906775914132595,
+      "num_tokens": 180295550.0,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_math_entropy": 0.182483324944973,
+      "eval_math_loss": 1.3162635564804077,
+      "eval_math_mean_token_accuracy": 0.8159587254524231,
+      "eval_math_num_tokens": 180295550.0,
+      "eval_math_runtime": 35.7935,
+      "eval_math_samples_per_second": 13.969,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6300
+    },
+    {
+      "epoch": 10.08,
+      "eval_physics_entropy": 0.20883261638879777,
+      "eval_physics_loss": 1.385360836982727,
+      "eval_physics_mean_token_accuracy": 0.7965086388587952,
+      "eval_physics_num_tokens": 180295550.0,
+      "eval_physics_runtime": 41.4527,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6300
+    },
+    {
+      "entropy": 0.05292724183527753,
+      "epoch": 10.096,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.202222222222222e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905779302120209,
+      "num_tokens": 180583471.0,
+      "step": 6310
+    },
+    {
+      "entropy": 0.051366113941185176,
+      "epoch": 10.112,
+      "grad_norm": 0.408203125,
+      "learning_rate": 8.18e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909560974687338,
+      "num_tokens": 180874563.0,
+      "step": 6320
+    },
+    {
+      "entropy": 0.05223900515120476,
+      "epoch": 10.128,
+      "grad_norm": 0.375,
+      "learning_rate": 8.157777777777779e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9908134344965219,
+      "num_tokens": 181163276.0,
+      "step": 6330
+    },
+    {
+      "entropy": 0.05358587969094515,
+      "epoch": 10.144,
+      "grad_norm": 0.5859375,
+      "learning_rate": 8.135555555555557e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.990205043554306,
+      "num_tokens": 181443334.0,
+      "step": 6340
+    },
+    {
+      "entropy": 0.052495576441287994,
+      "epoch": 10.16,
+      "grad_norm": 0.3671875,
+      "learning_rate": 8.113333333333334e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9907550234347582,
+      "num_tokens": 181728567.0,
+      "step": 6350
+    },
+    {
+      "entropy": 0.053296228067483756,
+      "epoch": 10.176,
+      "grad_norm": 0.451171875,
+      "learning_rate": 8.091111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904772769659758,
+      "num_tokens": 182013165.0,
+      "step": 6360
+    },
+    {
+      "entropy": 0.05225419225171209,
+      "epoch": 10.192,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.06888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990550859645009,
+      "num_tokens": 182295532.0,
+      "step": 6370
+    },
+    {
+      "entropy": 0.05398479872383177,
+      "epoch": 10.208,
+      "grad_norm": 0.45703125,
+      "learning_rate": 8.046666666666667e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990156552568078,
+      "num_tokens": 182579186.0,
+      "step": 6380
+    },
+    {
+      "entropy": 0.05467651817016304,
+      "epoch": 10.224,
+      "grad_norm": 0.4609375,
+      "learning_rate": 8.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900331899523735,
+      "num_tokens": 182854053.0,
+      "step": 6390
+    },
+    {
+      "entropy": 0.052144162415061146,
+      "epoch": 10.24,
+      "grad_norm": 0.396484375,
+      "learning_rate": 8.002222222222222e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906157191842795,
+      "num_tokens": 183145164.0,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_math_entropy": 0.18178619807958604,
+      "eval_math_loss": 1.3181895017623901,
+      "eval_math_mean_token_accuracy": 0.8160892844200134,
+      "eval_math_num_tokens": 183145164.0,
+      "eval_math_runtime": 35.7842,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6400
+    },
+    {
+      "epoch": 10.24,
+      "eval_physics_entropy": 0.2083468474149704,
+      "eval_physics_loss": 1.3860576152801514,
+      "eval_physics_mean_token_accuracy": 0.7964692945480347,
+      "eval_physics_num_tokens": 183145164.0,
+      "eval_physics_runtime": 41.4517,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.016,
+      "step": 6400
+    },
+    {
+      "entropy": 0.05220753562171012,
+      "epoch": 10.256,
+      "grad_norm": 0.37890625,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9906813442707062,
+      "num_tokens": 183433571.0,
+      "step": 6410
+    },
+    {
+      "entropy": 0.052020283648744226,
+      "epoch": 10.272,
+      "grad_norm": 0.4296875,
+      "learning_rate": 7.957777777777779e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905898869037628,
+      "num_tokens": 183726359.0,
+      "step": 6420
+    },
+    {
+      "entropy": 0.052133556816261264,
+      "epoch": 10.288,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.935555555555557e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905950725078583,
+      "num_tokens": 184012138.0,
+      "step": 6430
+    },
+    {
+      "entropy": 0.05153557150624692,
+      "epoch": 10.304,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.913333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906015176326036,
+      "num_tokens": 184308185.0,
+      "step": 6440
+    },
+    {
+      "entropy": 0.05174093886744231,
+      "epoch": 10.32,
+      "grad_norm": 0.470703125,
+      "learning_rate": 7.891111111111112e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906507469713688,
+      "num_tokens": 184598642.0,
+      "step": 6450
+    },
+    {
+      "entropy": 0.05283462274819613,
+      "epoch": 10.336,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.86888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904411014169454,
+      "num_tokens": 184889238.0,
+      "step": 6460
+    },
+    {
+      "entropy": 0.05393118744250387,
+      "epoch": 10.352,
+      "grad_norm": 0.412109375,
+      "learning_rate": 7.846666666666667e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903161883354187,
+      "num_tokens": 185174063.0,
+      "step": 6470
+    },
+    {
+      "entropy": 0.05220061105210334,
+      "epoch": 10.368,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.824444444444445e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904733497649432,
+      "num_tokens": 185463841.0,
+      "step": 6480
+    },
+    {
+      "entropy": 0.05435825029853732,
+      "epoch": 10.384,
+      "grad_norm": 0.44921875,
+      "learning_rate": 7.802222222222222e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901794776320457,
+      "num_tokens": 185743666.0,
+      "step": 6490
+    },
+    {
+      "entropy": 0.053181433398276565,
+      "epoch": 10.4,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.78e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.990223279595375,
+      "num_tokens": 186025983.0,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_math_entropy": 0.18081647664308548,
+      "eval_math_loss": 1.3202934265136719,
+      "eval_math_mean_token_accuracy": 0.8161701111793518,
+      "eval_math_num_tokens": 186025983.0,
+      "eval_math_runtime": 35.7672,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6500
+    },
+    {
+      "epoch": 10.4,
+      "eval_physics_entropy": 0.20749859195947648,
+      "eval_physics_loss": 1.3875044584274292,
+      "eval_physics_mean_token_accuracy": 0.7965818634033203,
+      "eval_physics_num_tokens": 186025983.0,
+      "eval_physics_runtime": 41.1812,
+      "eval_physics_samples_per_second": 12.141,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 6500
+    },
+    {
+      "entropy": 0.05349606759846211,
+      "epoch": 10.416,
+      "grad_norm": 0.4375,
+      "learning_rate": 7.75777777777778e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900016512721777,
+      "num_tokens": 186307901.0,
+      "step": 6510
+    },
+    {
+      "entropy": 0.05136881017824635,
+      "epoch": 10.432,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.735555555555557e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906102675944567,
+      "num_tokens": 186599883.0,
+      "step": 6520
+    },
+    {
+      "entropy": 0.053340693842619656,
+      "epoch": 10.448,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.713333333333335e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902616281062364,
+      "num_tokens": 186881626.0,
+      "step": 6530
+    },
+    {
+      "entropy": 0.05143804214894772,
+      "epoch": 10.464,
+      "grad_norm": 0.43359375,
+      "learning_rate": 7.691111111111112e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9908057555556298,
+      "num_tokens": 187174275.0,
+      "step": 6540
+    },
+    {
+      "entropy": 0.05378174295183271,
+      "epoch": 10.48,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.66888888888889e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901928540319205,
+      "num_tokens": 187457125.0,
+      "step": 6550
+    },
+    {
+      "entropy": 0.052962092636153105,
+      "epoch": 10.496,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.646666666666667e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902697995305061,
+      "num_tokens": 187745617.0,
+      "step": 6560
+    },
+    {
+      "entropy": 0.052524275076575574,
+      "epoch": 10.512,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.624444444444445e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902585811913014,
+      "num_tokens": 188028367.0,
+      "step": 6570
+    },
+    {
+      "entropy": 0.05309811832848936,
+      "epoch": 10.528,
+      "grad_norm": 0.373046875,
+      "learning_rate": 7.602222222222223e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902503415942192,
+      "num_tokens": 188313122.0,
+      "step": 6580
+    },
+    {
+      "entropy": 0.05307653601048514,
+      "epoch": 10.544,
+      "grad_norm": 0.423828125,
+      "learning_rate": 7.58e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9903011079877615,
+      "num_tokens": 188602441.0,
+      "step": 6590
+    },
+    {
+      "entropy": 0.05293632268439978,
+      "epoch": 10.56,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.557777777777779e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902526248246432,
+      "num_tokens": 188887324.0,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_math_entropy": 0.18184967929124832,
+      "eval_math_loss": 1.3195786476135254,
+      "eval_math_mean_token_accuracy": 0.8163261613845825,
+      "eval_math_num_tokens": 188887324.0,
+      "eval_math_runtime": 35.801,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 6600
+    },
+    {
+      "epoch": 10.56,
+      "eval_physics_entropy": 0.20828607833385468,
+      "eval_physics_loss": 1.3864657878875732,
+      "eval_physics_mean_token_accuracy": 0.7965416917800904,
+      "eval_physics_num_tokens": 188887324.0,
+      "eval_physics_runtime": 41.4621,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6600
+    },
+    {
+      "entropy": 0.05260960457380861,
+      "epoch": 10.576,
+      "grad_norm": 0.400390625,
+      "learning_rate": 7.535555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990390632674098,
+      "num_tokens": 189177342.0,
+      "step": 6610
+    },
+    {
+      "entropy": 0.05413947924971581,
+      "epoch": 10.592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.513333333333334e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9901154443621636,
+      "num_tokens": 189456532.0,
+      "step": 6620
+    },
+    {
+      "entropy": 0.05437520539853722,
+      "epoch": 10.608,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.4911111111111116e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900374330580235,
+      "num_tokens": 189739907.0,
+      "step": 6630
+    },
+    {
+      "entropy": 0.05323726618662476,
+      "epoch": 10.624,
+      "grad_norm": 0.384765625,
+      "learning_rate": 7.46888888888889e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904917422682047,
+      "num_tokens": 190028804.0,
+      "step": 6640
+    },
+    {
+      "entropy": 0.05413167366059497,
+      "epoch": 10.64,
+      "grad_norm": 0.41015625,
+      "learning_rate": 7.446666666666668e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900175202637911,
+      "num_tokens": 190313057.0,
+      "step": 6650
+    },
+    {
+      "entropy": 0.05292349813971668,
+      "epoch": 10.656,
+      "grad_norm": 0.392578125,
+      "learning_rate": 7.424444444444445e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903775095939636,
+      "num_tokens": 190601609.0,
+      "step": 6660
+    },
+    {
+      "entropy": 0.05083155656466261,
+      "epoch": 10.672,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.402222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9907138582319022,
+      "num_tokens": 190893252.0,
+      "step": 6670
+    },
+    {
+      "entropy": 0.05505718474742025,
+      "epoch": 10.688,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 0.0331,
+      "mean_token_accuracy": 0.9898338593542576,
+      "num_tokens": 191171634.0,
+      "step": 6680
+    },
+    {
+      "entropy": 0.0527726071421057,
+      "epoch": 10.704,
+      "grad_norm": 0.431640625,
+      "learning_rate": 7.357777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903190910816193,
+      "num_tokens": 191460983.0,
+      "step": 6690
+    },
+    {
+      "entropy": 0.05330016176449135,
+      "epoch": 10.72,
+      "grad_norm": 0.453125,
+      "learning_rate": 7.335555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901380114257335,
+      "num_tokens": 191748385.0,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_math_entropy": 0.18104988729953767,
+      "eval_math_loss": 1.3196566104888916,
+      "eval_math_mean_token_accuracy": 0.8163433685302734,
+      "eval_math_num_tokens": 191748385.0,
+      "eval_math_runtime": 35.7862,
+      "eval_math_samples_per_second": 13.972,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6700
+    },
+    {
+      "epoch": 10.72,
+      "eval_physics_entropy": 0.20756823128461838,
+      "eval_physics_loss": 1.3874235153198242,
+      "eval_physics_mean_token_accuracy": 0.7966196894645691,
+      "eval_physics_num_tokens": 191748385.0,
+      "eval_physics_runtime": 41.4894,
+      "eval_physics_samples_per_second": 12.051,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 6700
+    },
+    {
+      "entropy": 0.05236076007131487,
+      "epoch": 10.736,
+      "grad_norm": 0.451171875,
+      "learning_rate": 7.313333333333333e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902728144079447,
+      "num_tokens": 192036129.0,
+      "step": 6710
+    },
+    {
+      "entropy": 0.05269174930872396,
+      "epoch": 10.752,
+      "grad_norm": 0.408203125,
+      "learning_rate": 7.291111111111112e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990266852453351,
+      "num_tokens": 192323069.0,
+      "step": 6720
+    },
+    {
+      "entropy": 0.05275348348077387,
+      "epoch": 10.768,
+      "grad_norm": 0.416015625,
+      "learning_rate": 7.268888888888889e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903333097696304,
+      "num_tokens": 192610172.0,
+      "step": 6730
+    },
+    {
+      "entropy": 0.05378551658941433,
+      "epoch": 10.784,
+      "grad_norm": 0.421875,
+      "learning_rate": 7.246666666666667e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9902389969676733,
+      "num_tokens": 192892123.0,
+      "step": 6740
+    },
+    {
+      "entropy": 0.05370358480140567,
+      "epoch": 10.8,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.224444444444445e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9902192387729883,
+      "num_tokens": 193177806.0,
+      "step": 6750
+    },
+    {
+      "entropy": 0.05249750132206828,
+      "epoch": 10.816,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.202222222222222e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9903962533921004,
+      "num_tokens": 193465237.0,
+      "step": 6760
+    },
+    {
+      "entropy": 0.05406251410022378,
+      "epoch": 10.832,
+      "grad_norm": 0.36328125,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901404179632664,
+      "num_tokens": 193750065.0,
+      "step": 6770
+    },
+    {
+      "entropy": 0.05291999091859907,
+      "epoch": 10.848,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.157777777777778e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.99024215079844,
+      "num_tokens": 194039954.0,
+      "step": 6780
+    },
+    {
+      "entropy": 0.05192763227969408,
+      "epoch": 10.864,
+      "grad_norm": 0.42578125,
+      "learning_rate": 7.135555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904397703707218,
+      "num_tokens": 194328398.0,
+      "step": 6790
+    },
+    {
+      "entropy": 0.05234395726583898,
+      "epoch": 10.88,
+      "grad_norm": 0.40234375,
+      "learning_rate": 7.113333333333334e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902211964130402,
+      "num_tokens": 194616584.0,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_math_entropy": 0.1821578356027603,
+      "eval_math_loss": 1.3202202320098877,
+      "eval_math_mean_token_accuracy": 0.8162348837852478,
+      "eval_math_num_tokens": 194616584.0,
+      "eval_math_runtime": 35.7821,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 6800
+    },
+    {
+      "epoch": 10.88,
+      "eval_physics_entropy": 0.20837496918439866,
+      "eval_physics_loss": 1.3878107070922852,
+      "eval_physics_mean_token_accuracy": 0.7964894032478332,
+      "eval_physics_num_tokens": 194616584.0,
+      "eval_physics_runtime": 41.4623,
+      "eval_physics_samples_per_second": 12.059,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 6800
+    },
+    {
+      "entropy": 0.053814200405031445,
+      "epoch": 10.896,
+      "grad_norm": 0.41796875,
+      "learning_rate": 7.091111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9900967665016651,
+      "num_tokens": 194902364.0,
+      "step": 6810
+    },
+    {
+      "entropy": 0.05345905120484531,
+      "epoch": 10.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 7.06888888888889e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.9903675314038992,
+      "num_tokens": 195188102.0,
+      "step": 6820
+    },
+    {
+      "entropy": 0.05543638013768941,
+      "epoch": 10.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 7.046666666666667e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9897865749895572,
+      "num_tokens": 195466422.0,
+      "step": 6830
+    },
+    {
+      "entropy": 0.053550386987626555,
+      "epoch": 10.943999999999999,
+      "grad_norm": 0.380859375,
+      "learning_rate": 7.024444444444445e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900572828948497,
+      "num_tokens": 195748476.0,
+      "step": 6840
+    },
+    {
+      "entropy": 0.052956846286542715,
+      "epoch": 10.96,
+      "grad_norm": 0.404296875,
+      "learning_rate": 7.0022222222222225e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902277212589979,
+      "num_tokens": 196037944.0,
+      "step": 6850
+    },
+    {
+      "entropy": 0.053696213918738064,
+      "epoch": 10.975999999999999,
+      "grad_norm": 0.46875,
+      "learning_rate": 6.98e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900156583636999,
+      "num_tokens": 196322679.0,
+      "step": 6860
+    },
+    {
+      "entropy": 0.0527949180919677,
+      "epoch": 10.992,
+      "grad_norm": 0.41015625,
+      "learning_rate": 6.9577777777777785e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902477405965329,
+      "num_tokens": 196609593.0,
+      "step": 6870
+    },
+    {
+      "entropy": 0.05277393463766202,
+      "epoch": 11.008,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903519790619612,
+      "num_tokens": 196895126.0,
+      "step": 6880
+    },
+    {
+      "entropy": 0.050324175844434646,
+      "epoch": 11.024,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.913333333333334e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.991166714578867,
+      "num_tokens": 197194760.0,
+      "step": 6890
+    },
+    {
+      "entropy": 0.05295788950752467,
+      "epoch": 11.04,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.891111111111111e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906532682478428,
+      "num_tokens": 197480211.0,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_math_entropy": 0.1807521269917488,
+      "eval_math_loss": 1.3240716457366943,
+      "eval_math_mean_token_accuracy": 0.8160151686668397,
+      "eval_math_num_tokens": 197480211.0,
+      "eval_math_runtime": 35.764,
+      "eval_math_samples_per_second": 13.981,
+      "eval_math_steps_per_second": 3.495,
+      "step": 6900
+    },
+    {
+      "epoch": 11.04,
+      "eval_physics_entropy": 0.20740420019626618,
+      "eval_physics_loss": 1.3912010192871094,
+      "eval_physics_mean_token_accuracy": 0.7964357166290283,
+      "eval_physics_num_tokens": 197480211.0,
+      "eval_physics_runtime": 41.494,
+      "eval_physics_samples_per_second": 12.05,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 6900
+    },
+    {
+      "entropy": 0.05364316294435412,
+      "epoch": 11.056,
+      "grad_norm": 0.4140625,
+      "learning_rate": 6.868888888888889e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903167489916086,
+      "num_tokens": 197756071.0,
+      "step": 6910
+    },
+    {
+      "entropy": 0.05119335462804884,
+      "epoch": 11.072,
+      "grad_norm": 0.40625,
+      "learning_rate": 6.846666666666667e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908623903989792,
+      "num_tokens": 198048655.0,
+      "step": 6920
+    },
+    {
+      "entropy": 0.05139227455947548,
+      "epoch": 11.088,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.824444444444444e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908630765974522,
+      "num_tokens": 198339840.0,
+      "step": 6930
+    },
+    {
+      "entropy": 0.052823137422092256,
+      "epoch": 11.104,
+      "grad_norm": 0.486328125,
+      "learning_rate": 6.802222222222223e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906187053769827,
+      "num_tokens": 198626746.0,
+      "step": 6940
+    },
+    {
+      "entropy": 0.05260417617391795,
+      "epoch": 11.12,
+      "grad_norm": 0.421875,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904017817229033,
+      "num_tokens": 198909663.0,
+      "step": 6950
+    },
+    {
+      "entropy": 0.05475040662568063,
+      "epoch": 11.136,
+      "grad_norm": 0.458984375,
+      "learning_rate": 6.757777777777779e-06,
+      "loss": 0.0335,
+      "mean_token_accuracy": 0.9898043904453516,
+      "num_tokens": 199178288.0,
+      "step": 6960
+    },
+    {
+      "entropy": 0.05385386899579316,
+      "epoch": 11.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 6.735555555555556e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9901543207466602,
+      "num_tokens": 199455295.0,
+      "step": 6970
+    },
+    {
+      "entropy": 0.05362422023899853,
+      "epoch": 11.168,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.713333333333334e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9903968300670385,
+      "num_tokens": 199730601.0,
+      "step": 6980
+    },
+    {
+      "entropy": 0.05283083380199969,
+      "epoch": 11.184,
+      "grad_norm": 0.431640625,
+      "learning_rate": 6.691111111111112e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9904793795198202,
+      "num_tokens": 200012276.0,
+      "step": 6990
+    },
+    {
+      "entropy": 0.05167927134316415,
+      "epoch": 11.2,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.668888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905506029725075,
+      "num_tokens": 200302889.0,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_math_entropy": 0.1808690933585167,
+      "eval_math_loss": 1.328149437904358,
+      "eval_math_mean_token_accuracy": 0.8160120005607605,
+      "eval_math_num_tokens": 200302889.0,
+      "eval_math_runtime": 35.7829,
+      "eval_math_samples_per_second": 13.973,
+      "eval_math_steps_per_second": 3.493,
+      "step": 7000
+    },
+    {
+      "epoch": 11.2,
+      "eval_physics_entropy": 0.20758735692501068,
+      "eval_physics_loss": 1.3947290182113647,
+      "eval_physics_mean_token_accuracy": 0.7964826035499573,
+      "eval_physics_num_tokens": 200302889.0,
+      "eval_physics_runtime": 41.1925,
+      "eval_physics_samples_per_second": 12.138,
+      "eval_physics_steps_per_second": 3.035,
+      "step": 7000
+    },
+    {
+      "entropy": 0.05169075320009142,
+      "epoch": 11.216,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.646666666666667e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907121449708939,
+      "num_tokens": 200594123.0,
+      "step": 7010
+    },
+    {
+      "entropy": 0.05046803302830085,
+      "epoch": 11.232,
+      "grad_norm": 0.390625,
+      "learning_rate": 6.6244444444444445e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990829760953784,
+      "num_tokens": 200883523.0,
+      "step": 7020
+    },
+    {
+      "entropy": 0.05166568392887712,
+      "epoch": 11.248,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.602222222222223e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904365949332714,
+      "num_tokens": 201170610.0,
+      "step": 7030
+    },
+    {
+      "entropy": 0.052717051294166595,
+      "epoch": 11.264,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9901825904846191,
+      "num_tokens": 201455947.0,
+      "step": 7040
+    },
+    {
+      "entropy": 0.053193623945116994,
+      "epoch": 11.28,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9901075243949891,
+      "num_tokens": 201736602.0,
+      "step": 7050
+    },
+    {
+      "entropy": 0.052154396660625935,
+      "epoch": 11.296,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.535555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9904614068567753,
+      "num_tokens": 202016426.0,
+      "step": 7060
+    },
+    {
+      "entropy": 0.05196985362563282,
+      "epoch": 11.312,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.513333333333333e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905673917382956,
+      "num_tokens": 202304355.0,
+      "step": 7070
+    },
+    {
+      "entropy": 0.05061813323991373,
+      "epoch": 11.328,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.491111111111111e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9909276533871889,
+      "num_tokens": 202596962.0,
+      "step": 7080
+    },
+    {
+      "entropy": 0.05172014329582453,
+      "epoch": 11.344,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904945332556963,
+      "num_tokens": 202886289.0,
+      "step": 7090
+    },
+    {
+      "entropy": 0.0515604832675308,
+      "epoch": 11.36,
+      "grad_norm": 0.423828125,
+      "learning_rate": 6.446666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905925836414099,
+      "num_tokens": 203176925.0,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_math_entropy": 0.18083494365215302,
+      "eval_math_loss": 1.3280311822891235,
+      "eval_math_mean_token_accuracy": 0.8161394634246826,
+      "eval_math_num_tokens": 203176925.0,
+      "eval_math_runtime": 35.8601,
+      "eval_math_samples_per_second": 13.943,
+      "eval_math_steps_per_second": 3.486,
+      "step": 7100
+    },
+    {
+      "epoch": 11.36,
+      "eval_physics_entropy": 0.20746298998594284,
+      "eval_physics_loss": 1.3944538831710815,
+      "eval_physics_mean_token_accuracy": 0.7962424664497375,
+      "eval_physics_num_tokens": 203176925.0,
+      "eval_physics_runtime": 41.5618,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7100
+    },
+    {
+      "entropy": 0.05180059259291738,
+      "epoch": 11.376,
+      "grad_norm": 0.451171875,
+      "learning_rate": 6.4244444444444455e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990391731634736,
+      "num_tokens": 203460548.0,
+      "step": 7110
+    },
+    {
+      "entropy": 0.0532453742576763,
+      "epoch": 11.392,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.402222222222223e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9901927445083857,
+      "num_tokens": 203740569.0,
+      "step": 7120
+    },
+    {
+      "entropy": 0.05357764135114849,
+      "epoch": 11.408,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 0.0329,
+      "mean_token_accuracy": 0.9900757193565368,
+      "num_tokens": 204015895.0,
+      "step": 7130
+    },
+    {
+      "entropy": 0.05195869819726795,
+      "epoch": 11.424,
+      "grad_norm": 0.453125,
+      "learning_rate": 6.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903892528265714,
+      "num_tokens": 204303156.0,
+      "step": 7140
+    },
+    {
+      "entropy": 0.0533925520721823,
+      "epoch": 11.44,
+      "grad_norm": 0.373046875,
+      "learning_rate": 6.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903348978608847,
+      "num_tokens": 204584978.0,
+      "step": 7150
+    },
+    {
+      "entropy": 0.0531626753276214,
+      "epoch": 11.456,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.313333333333334e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904294617474079,
+      "num_tokens": 204870321.0,
+      "step": 7160
+    },
+    {
+      "entropy": 0.051605701842345295,
+      "epoch": 11.472,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.291111111111111e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906095393002033,
+      "num_tokens": 205157148.0,
+      "step": 7170
+    },
+    {
+      "entropy": 0.051634928002022205,
+      "epoch": 11.488,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.26888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9907478917390108,
+      "num_tokens": 205447025.0,
+      "step": 7180
+    },
+    {
+      "entropy": 0.050146938033867626,
+      "epoch": 11.504,
+      "grad_norm": 0.4296875,
+      "learning_rate": 6.246666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9909087587147951,
+      "num_tokens": 205743272.0,
+      "step": 7190
+    },
+    {
+      "entropy": 0.05206783713074401,
+      "epoch": 11.52,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.224444444444445e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905533608049154,
+      "num_tokens": 206031966.0,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_math_entropy": 0.18023463946580887,
+      "eval_math_loss": 1.3275227546691895,
+      "eval_math_mean_token_accuracy": 0.8161748089790344,
+      "eval_math_num_tokens": 206031966.0,
+      "eval_math_runtime": 35.8477,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 7200
+    },
+    {
+      "epoch": 11.52,
+      "eval_physics_entropy": 0.20684081387519837,
+      "eval_physics_loss": 1.3938384056091309,
+      "eval_physics_mean_token_accuracy": 0.7965911884307861,
+      "eval_physics_num_tokens": 206031966.0,
+      "eval_physics_runtime": 41.5446,
+      "eval_physics_samples_per_second": 12.035,
+      "eval_physics_steps_per_second": 3.009,
+      "step": 7200
+    },
+    {
+      "entropy": 0.05208216350292787,
+      "epoch": 11.536,
+      "grad_norm": 0.447265625,
+      "learning_rate": 6.2022222222222225e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9901717752218246,
+      "num_tokens": 206315869.0,
+      "step": 7210
+    },
+    {
+      "entropy": 0.052281872206367555,
+      "epoch": 11.552,
+      "grad_norm": 0.416015625,
+      "learning_rate": 6.18e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904672216624022,
+      "num_tokens": 206602917.0,
+      "step": 7220
+    },
+    {
+      "entropy": 0.05279321176931262,
+      "epoch": 11.568,
+      "grad_norm": 0.435546875,
+      "learning_rate": 6.157777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9903134804219007,
+      "num_tokens": 206884919.0,
+      "step": 7230
+    },
+    {
+      "entropy": 0.052229331363923846,
+      "epoch": 11.584,
+      "grad_norm": 0.4453125,
+      "learning_rate": 6.135555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903485018759965,
+      "num_tokens": 207170989.0,
+      "step": 7240
+    },
+    {
+      "entropy": 0.052299269544892016,
+      "epoch": 11.6,
+      "grad_norm": 0.4375,
+      "learning_rate": 6.113333333333333e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903657089918851,
+      "num_tokens": 207459707.0,
+      "step": 7250
+    },
+    {
+      "entropy": 0.05078096168581396,
+      "epoch": 11.616,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.091111111111112e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908219832926989,
+      "num_tokens": 207755276.0,
+      "step": 7260
+    },
+    {
+      "entropy": 0.05125759767834097,
+      "epoch": 11.632,
+      "grad_norm": 0.388671875,
+      "learning_rate": 6.06888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903901822865009,
+      "num_tokens": 208045204.0,
+      "step": 7270
+    },
+    {
+      "entropy": 0.05171195106813684,
+      "epoch": 11.648,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.0466666666666675e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904062520712614,
+      "num_tokens": 208331207.0,
+      "step": 7280
+    },
+    {
+      "entropy": 0.05101303905248642,
+      "epoch": 11.664,
+      "grad_norm": 0.44140625,
+      "learning_rate": 6.024444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906638719141483,
+      "num_tokens": 208625921.0,
+      "step": 7290
+    },
+    {
+      "entropy": 0.05110592239070684,
+      "epoch": 11.68,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.002222222222223e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907457642257214,
+      "num_tokens": 208923187.0,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_math_entropy": 0.180965376496315,
+      "eval_math_loss": 1.3272041082382202,
+      "eval_math_mean_token_accuracy": 0.8161662096977234,
+      "eval_math_num_tokens": 208923187.0,
+      "eval_math_runtime": 35.8661,
+      "eval_math_samples_per_second": 13.941,
+      "eval_math_steps_per_second": 3.485,
+      "step": 7300
+    },
+    {
+      "epoch": 11.68,
+      "eval_physics_entropy": 0.20749008816480635,
+      "eval_physics_loss": 1.393621802330017,
+      "eval_physics_mean_token_accuracy": 0.7964129576683044,
+      "eval_physics_num_tokens": 208923187.0,
+      "eval_physics_runtime": 41.5619,
+      "eval_physics_samples_per_second": 12.03,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 7300
+    },
+    {
+      "entropy": 0.05148296125698835,
+      "epoch": 11.696,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.98e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905769471079111,
+      "num_tokens": 209212129.0,
+      "step": 7310
+    },
+    {
+      "entropy": 0.052077129227109255,
+      "epoch": 11.712,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.957777777777778e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904128357768058,
+      "num_tokens": 209504158.0,
+      "step": 7320
+    },
+    {
+      "entropy": 0.052831382700242104,
+      "epoch": 11.728,
+      "grad_norm": 0.412109375,
+      "learning_rate": 5.9355555555555556e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902264326810837,
+      "num_tokens": 209787517.0,
+      "step": 7330
+    },
+    {
+      "entropy": 0.051228555967099965,
+      "epoch": 11.744,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.913333333333334e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906369145959616,
+      "num_tokens": 210080720.0,
+      "step": 7340
+    },
+    {
+      "entropy": 0.05308413691818714,
+      "epoch": 11.76,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.891111111111112e-06,
+      "loss": 0.0325,
+      "mean_token_accuracy": 0.9902846306562424,
+      "num_tokens": 210361593.0,
+      "step": 7350
+    },
+    {
+      "entropy": 0.05108139000367373,
+      "epoch": 11.776,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.868888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990474434569478,
+      "num_tokens": 210657639.0,
+      "step": 7360
+    },
+    {
+      "entropy": 0.052184423850849274,
+      "epoch": 11.792,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.846666666666667e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9901965599507093,
+      "num_tokens": 210946084.0,
+      "step": 7370
+    },
+    {
+      "entropy": 0.05307481104973703,
+      "epoch": 11.808,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.8244444444444445e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.9902853887528181,
+      "num_tokens": 211230941.0,
+      "step": 7380
+    },
+    {
+      "entropy": 0.05265059581724927,
+      "epoch": 11.824,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.802222222222222e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903722569346428,
+      "num_tokens": 211518349.0,
+      "step": 7390
+    },
+    {
+      "entropy": 0.051535707782022655,
+      "epoch": 11.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 5.78e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990499372035265,
+      "num_tokens": 211813189.0,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_math_entropy": 0.18025576132535934,
+      "eval_math_loss": 1.3284978866577148,
+      "eval_math_mean_token_accuracy": 0.8162349495887756,
+      "eval_math_num_tokens": 211813189.0,
+      "eval_math_runtime": 36.1844,
+      "eval_math_samples_per_second": 13.818,
+      "eval_math_steps_per_second": 3.455,
+      "step": 7400
+    },
+    {
+      "epoch": 11.84,
+      "eval_physics_entropy": 0.20679952079057692,
+      "eval_physics_loss": 1.3946645259857178,
+      "eval_physics_mean_token_accuracy": 0.7964466457366943,
+      "eval_physics_num_tokens": 211813189.0,
+      "eval_physics_runtime": 41.5186,
+      "eval_physics_samples_per_second": 12.043,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7400
+    },
+    {
+      "entropy": 0.05275586240459233,
+      "epoch": 11.856,
+      "grad_norm": 0.46875,
+      "learning_rate": 5.757777777777779e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899599272757769,
+      "num_tokens": 212093999.0,
+      "step": 7410
+    },
+    {
+      "entropy": 0.052845566172618416,
+      "epoch": 11.872,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.735555555555557e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990190564095974,
+      "num_tokens": 212380039.0,
+      "step": 7420
+    },
+    {
+      "entropy": 0.05428367832209915,
+      "epoch": 11.888,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.713333333333334e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9899089831858874,
+      "num_tokens": 212663647.0,
+      "step": 7430
+    },
+    {
+      "entropy": 0.052614666032604875,
+      "epoch": 11.904,
+      "grad_norm": 0.419921875,
+      "learning_rate": 5.691111111111112e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.990277175605297,
+      "num_tokens": 212945404.0,
+      "step": 7440
+    },
+    {
+      "entropy": 0.053846529638394716,
+      "epoch": 11.92,
+      "grad_norm": 0.455078125,
+      "learning_rate": 5.6688888888888895e-06,
+      "loss": 0.0332,
+      "mean_token_accuracy": 0.989962762594223,
+      "num_tokens": 213220908.0,
+      "step": 7450
+    },
+    {
+      "entropy": 0.053387761081103234,
+      "epoch": 11.936,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.646666666666667e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9899429127573967,
+      "num_tokens": 213501339.0,
+      "step": 7460
+    },
+    {
+      "entropy": 0.05214737723581493,
+      "epoch": 11.952,
+      "grad_norm": 0.390625,
+      "learning_rate": 5.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9902004975825548,
+      "num_tokens": 213787969.0,
+      "step": 7470
+    },
+    {
+      "entropy": 0.05322346042376012,
+      "epoch": 11.968,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.602222222222222e-06,
+      "loss": 0.0328,
+      "mean_token_accuracy": 0.9901369668543338,
+      "num_tokens": 214065638.0,
+      "step": 7480
+    },
+    {
+      "entropy": 0.0515533285215497,
+      "epoch": 11.984,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9904691580682993,
+      "num_tokens": 214356584.0,
+      "step": 7490
+    },
+    {
+      "entropy": 0.05252209414029494,
+      "epoch": 12.0,
+      "grad_norm": 0.46484375,
+      "learning_rate": 5.557777777777778e-06,
+      "loss": 0.0327,
+      "mean_token_accuracy": 0.9901051189750433,
+      "num_tokens": 214637088.0,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_math_entropy": 0.18112814128398896,
+      "eval_math_loss": 1.3286715745925903,
+      "eval_math_mean_token_accuracy": 0.8161366939544678,
+      "eval_math_num_tokens": 214637088.0,
+      "eval_math_runtime": 35.8423,
+      "eval_math_samples_per_second": 13.95,
+      "eval_math_steps_per_second": 3.488,
+      "step": 7500
+    },
+    {
+      "epoch": 12.0,
+      "eval_physics_entropy": 0.20745954370498657,
+      "eval_physics_loss": 1.3952056169509888,
+      "eval_physics_mean_token_accuracy": 0.796566994190216,
+      "eval_physics_num_tokens": 214637088.0,
+      "eval_physics_runtime": 41.171,
+      "eval_physics_samples_per_second": 12.144,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 7500
+    },
+    {
+      "entropy": 0.053174392925575374,
+      "epoch": 12.016,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.535555555555556e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905574269592762,
+      "num_tokens": 214920834.0,
+      "step": 7510
+    },
+    {
+      "entropy": 0.05121934600174427,
+      "epoch": 12.032,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.513333333333334e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907024279236794,
+      "num_tokens": 215207212.0,
+      "step": 7520
+    },
+    {
+      "entropy": 0.05282459647860378,
+      "epoch": 12.048,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.491111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904284842312336,
+      "num_tokens": 215491551.0,
+      "step": 7530
+    },
+    {
+      "entropy": 0.0523437969153747,
+      "epoch": 12.064,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.468888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9907445590943098,
+      "num_tokens": 215773788.0,
+      "step": 7540
+    },
+    {
+      "entropy": 0.05109455803176388,
+      "epoch": 12.08,
+      "grad_norm": 0.439453125,
+      "learning_rate": 5.4466666666666665e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990898734331131,
+      "num_tokens": 216063342.0,
+      "step": 7550
+    },
+    {
+      "entropy": 0.052485697087831795,
+      "epoch": 12.096,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.424444444444444e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905819371342659,
+      "num_tokens": 216346341.0,
+      "step": 7560
+    },
+    {
+      "entropy": 0.05154973501339555,
+      "epoch": 12.112,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.402222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906920839101077,
+      "num_tokens": 216631586.0,
+      "step": 7570
+    },
+    {
+      "entropy": 0.05027051754295826,
+      "epoch": 12.128,
+      "grad_norm": 0.404296875,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908485073596239,
+      "num_tokens": 216921387.0,
+      "step": 7580
+    },
+    {
+      "entropy": 0.051271379156969485,
+      "epoch": 12.144,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.357777777777779e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9907938674092293,
+      "num_tokens": 217208948.0,
+      "step": 7590
+    },
+    {
+      "entropy": 0.05280535111669451,
+      "epoch": 12.16,
+      "grad_norm": 0.427734375,
+      "learning_rate": 5.335555555555556e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904177527874708,
+      "num_tokens": 217488087.0,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_math_entropy": 0.17994125121831894,
+      "eval_math_loss": 1.3322224617004395,
+      "eval_math_mean_token_accuracy": 0.8160332980155944,
+      "eval_math_num_tokens": 217488087.0,
+      "eval_math_runtime": 35.8103,
+      "eval_math_samples_per_second": 13.962,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7600
+    },
+    {
+      "epoch": 12.16,
+      "eval_physics_entropy": 0.20668217623233795,
+      "eval_physics_loss": 1.3987233638763428,
+      "eval_physics_mean_token_accuracy": 0.7963971982002258,
+      "eval_physics_num_tokens": 217488087.0,
+      "eval_physics_runtime": 41.4958,
+      "eval_physics_samples_per_second": 12.049,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7600
+    },
+    {
+      "entropy": 0.05196192401926965,
+      "epoch": 12.176,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.313333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9906202860176563,
+      "num_tokens": 217770038.0,
+      "step": 7610
+    },
+    {
+      "entropy": 0.051902253658045086,
+      "epoch": 12.192,
+      "grad_norm": 0.435546875,
+      "learning_rate": 5.2911111111111115e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9906028714030981,
+      "num_tokens": 218050486.0,
+      "step": 7620
+    },
+    {
+      "entropy": 0.052790193830151114,
+      "epoch": 12.208,
+      "grad_norm": 0.443359375,
+      "learning_rate": 5.268888888888889e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904781151562929,
+      "num_tokens": 218331465.0,
+      "step": 7630
+    },
+    {
+      "entropy": 0.052298348920885473,
+      "epoch": 12.224,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.246666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906631808727979,
+      "num_tokens": 218618801.0,
+      "step": 7640
+    },
+    {
+      "entropy": 0.050798607314936815,
+      "epoch": 12.24,
+      "grad_norm": 0.431640625,
+      "learning_rate": 5.224444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9909202449023724,
+      "num_tokens": 218910596.0,
+      "step": 7650
+    },
+    {
+      "entropy": 0.052114113536663356,
+      "epoch": 12.256,
+      "grad_norm": 0.416015625,
+      "learning_rate": 5.202222222222223e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905635304749012,
+      "num_tokens": 219194842.0,
+      "step": 7660
+    },
+    {
+      "entropy": 0.05205410345224663,
+      "epoch": 12.272,
+      "grad_norm": 0.41015625,
+      "learning_rate": 5.18e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990602058544755,
+      "num_tokens": 219484076.0,
+      "step": 7670
+    },
+    {
+      "entropy": 0.049987882829736915,
+      "epoch": 12.288,
+      "grad_norm": 0.396484375,
+      "learning_rate": 5.157777777777778e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909869834780693,
+      "num_tokens": 219780222.0,
+      "step": 7680
+    },
+    {
+      "entropy": 0.05258604409173131,
+      "epoch": 12.304,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.135555555555556e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904522281140089,
+      "num_tokens": 220065272.0,
+      "step": 7690
+    },
+    {
+      "entropy": 0.05357218148419633,
+      "epoch": 12.32,
+      "grad_norm": 0.4453125,
+      "learning_rate": 5.113333333333333e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9902280792593956,
+      "num_tokens": 220344110.0,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_math_entropy": 0.1793700819015503,
+      "eval_math_loss": 1.3329575061798096,
+      "eval_math_mean_token_accuracy": 0.8161038937568664,
+      "eval_math_num_tokens": 220344110.0,
+      "eval_math_runtime": 35.8174,
+      "eval_math_samples_per_second": 13.96,
+      "eval_math_steps_per_second": 3.49,
+      "step": 7700
+    },
+    {
+      "epoch": 12.32,
+      "eval_physics_entropy": 0.20628326386213303,
+      "eval_physics_loss": 1.3988101482391357,
+      "eval_physics_mean_token_accuracy": 0.7963225340843201,
+      "eval_physics_num_tokens": 220344110.0,
+      "eval_physics_runtime": 41.5047,
+      "eval_physics_samples_per_second": 12.047,
+      "eval_physics_steps_per_second": 3.012,
+      "step": 7700
+    },
+    {
+      "entropy": 0.05083432825049385,
+      "epoch": 12.336,
+      "grad_norm": 0.42578125,
+      "learning_rate": 5.091111111111111e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906401328742505,
+      "num_tokens": 220632516.0,
+      "step": 7710
+    },
+    {
+      "entropy": 0.05188987164292484,
+      "epoch": 12.352,
+      "grad_norm": 0.43359375,
+      "learning_rate": 5.06888888888889e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905499298125505,
+      "num_tokens": 220918123.0,
+      "step": 7720
+    },
+    {
+      "entropy": 0.05209730467759073,
+      "epoch": 12.368,
+      "grad_norm": 0.40625,
+      "learning_rate": 5.046666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905961699783802,
+      "num_tokens": 221206063.0,
+      "step": 7730
+    },
+    {
+      "entropy": 0.05238552400842309,
+      "epoch": 12.384,
+      "grad_norm": 0.41796875,
+      "learning_rate": 5.024444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904579646885395,
+      "num_tokens": 221488317.0,
+      "step": 7740
+    },
+    {
+      "entropy": 0.05181863873731345,
+      "epoch": 12.4,
+      "grad_norm": 0.4140625,
+      "learning_rate": 5.002222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905485171824694,
+      "num_tokens": 221776008.0,
+      "step": 7750
+    },
+    {
+      "entropy": 0.05192954689264297,
+      "epoch": 12.416,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904752410948277,
+      "num_tokens": 222057789.0,
+      "step": 7760
+    },
+    {
+      "entropy": 0.052794202556833626,
+      "epoch": 12.432,
+      "grad_norm": 0.4453125,
+      "learning_rate": 4.957777777777778e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990259325504303,
+      "num_tokens": 222339915.0,
+      "step": 7770
+    },
+    {
+      "entropy": 0.052391084167174994,
+      "epoch": 12.448,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.935555555555556e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9903523031622171,
+      "num_tokens": 222619687.0,
+      "step": 7780
+    },
+    {
+      "entropy": 0.05029154985677451,
+      "epoch": 12.464,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.9133333333333334e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906887255609036,
+      "num_tokens": 222913264.0,
+      "step": 7790
+    },
+    {
+      "entropy": 0.05082799810916185,
+      "epoch": 12.48,
+      "grad_norm": 0.44921875,
+      "learning_rate": 4.891111111111111e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906626105308532,
+      "num_tokens": 223202947.0,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_math_entropy": 0.180211255133152,
+      "eval_math_loss": 1.333121657371521,
+      "eval_math_mean_token_accuracy": 0.8160158872604371,
+      "eval_math_num_tokens": 223202947.0,
+      "eval_math_runtime": 35.8241,
+      "eval_math_samples_per_second": 13.957,
+      "eval_math_steps_per_second": 3.489,
+      "step": 7800
+    },
+    {
+      "epoch": 12.48,
+      "eval_physics_entropy": 0.20699969559907913,
+      "eval_physics_loss": 1.3986775875091553,
+      "eval_physics_mean_token_accuracy": 0.7963728904724121,
+      "eval_physics_num_tokens": 223202947.0,
+      "eval_physics_runtime": 41.4724,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 7800
+    },
+    {
+      "entropy": 0.05148934966418892,
+      "epoch": 12.496,
+      "grad_norm": 0.48046875,
+      "learning_rate": 4.8688888888888895e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905773665755987,
+      "num_tokens": 223491040.0,
+      "step": 7810
+    },
+    {
+      "entropy": 0.05243352035759017,
+      "epoch": 12.512,
+      "grad_norm": 0.515625,
+      "learning_rate": 4.846666666666667e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903561118990183,
+      "num_tokens": 223775825.0,
+      "step": 7820
+    },
+    {
+      "entropy": 0.05058162130881101,
+      "epoch": 12.528,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.824444444444445e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990830871462822,
+      "num_tokens": 224068092.0,
+      "step": 7830
+    },
+    {
+      "entropy": 0.051722967089153825,
+      "epoch": 12.544,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.802222222222222e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903257399797439,
+      "num_tokens": 224354624.0,
+      "step": 7840
+    },
+    {
+      "entropy": 0.05123099538031965,
+      "epoch": 12.56,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.78e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9906783111393451,
+      "num_tokens": 224641974.0,
+      "step": 7850
+    },
+    {
+      "entropy": 0.05136660183779895,
+      "epoch": 12.576,
+      "grad_norm": 0.51171875,
+      "learning_rate": 4.7577777777777784e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906629983335733,
+      "num_tokens": 224932014.0,
+      "step": 7860
+    },
+    {
+      "entropy": 0.05145444863010198,
+      "epoch": 12.592,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.735555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9903172593563795,
+      "num_tokens": 225222434.0,
+      "step": 7870
+    },
+    {
+      "entropy": 0.05075638788985089,
+      "epoch": 12.608,
+      "grad_norm": 0.48828125,
+      "learning_rate": 4.713333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906791742891073,
+      "num_tokens": 225515803.0,
+      "step": 7880
+    },
+    {
+      "entropy": 0.05076117676217109,
+      "epoch": 12.624,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.691111111111111e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9906085334718228,
+      "num_tokens": 225810131.0,
+      "step": 7890
+    },
+    {
+      "entropy": 0.05141493146074936,
+      "epoch": 12.64,
+      "grad_norm": 0.40234375,
+      "learning_rate": 4.66888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904979523271322,
+      "num_tokens": 226100172.0,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_math_entropy": 0.17927622705698013,
+      "eval_math_loss": 1.332814335823059,
+      "eval_math_mean_token_accuracy": 0.8159863357543945,
+      "eval_math_num_tokens": 226100172.0,
+      "eval_math_runtime": 35.8058,
+      "eval_math_samples_per_second": 13.964,
+      "eval_math_steps_per_second": 3.491,
+      "step": 7900
+    },
+    {
+      "epoch": 12.64,
+      "eval_physics_entropy": 0.20605486339330673,
+      "eval_physics_loss": 1.3984746932983398,
+      "eval_physics_mean_token_accuracy": 0.7964399604797363,
+      "eval_physics_num_tokens": 226100172.0,
+      "eval_physics_runtime": 41.5077,
+      "eval_physics_samples_per_second": 12.046,
+      "eval_physics_steps_per_second": 3.011,
+      "step": 7900
+    },
+    {
+      "entropy": 0.052684175770264116,
+      "epoch": 12.656,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.646666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9902132380753755,
+      "num_tokens": 226379811.0,
+      "step": 7910
+    },
+    {
+      "entropy": 0.05348572533112019,
+      "epoch": 12.672,
+      "grad_norm": 0.470703125,
+      "learning_rate": 4.624444444444445e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.990258751064539,
+      "num_tokens": 226662085.0,
+      "step": 7920
+    },
+    {
+      "entropy": 0.05106104960432276,
+      "epoch": 12.688,
+      "grad_norm": 0.412109375,
+      "learning_rate": 4.602222222222223e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9902874868363142,
+      "num_tokens": 226950257.0,
+      "step": 7930
+    },
+    {
+      "entropy": 0.0518765551969409,
+      "epoch": 12.704,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905543696135283,
+      "num_tokens": 227238669.0,
+      "step": 7940
+    },
+    {
+      "entropy": 0.05253795627504587,
+      "epoch": 12.72,
+      "grad_norm": 0.38671875,
+      "learning_rate": 4.557777777777778e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990134946629405,
+      "num_tokens": 227518695.0,
+      "step": 7950
+    },
+    {
+      "entropy": 0.05207758563337848,
+      "epoch": 12.736,
+      "grad_norm": 0.43359375,
+      "learning_rate": 4.535555555555555e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990263070166111,
+      "num_tokens": 227806156.0,
+      "step": 7960
+    },
+    {
+      "entropy": 0.05179664236493409,
+      "epoch": 12.752,
+      "grad_norm": 0.443359375,
+      "learning_rate": 4.513333333333333e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904587935656309,
+      "num_tokens": 228091641.0,
+      "step": 7970
+    },
+    {
+      "entropy": 0.05190181992948055,
+      "epoch": 12.768,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.4911111111111115e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9901351545006036,
+      "num_tokens": 228376952.0,
+      "step": 7980
+    },
+    {
+      "entropy": 0.0528141722548753,
+      "epoch": 12.784,
+      "grad_norm": 0.47265625,
+      "learning_rate": 4.468888888888889e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901187218725681,
+      "num_tokens": 228657003.0,
+      "step": 7990
+    },
+    {
+      "entropy": 0.051391151722054926,
+      "epoch": 12.8,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.446666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9902074325829744,
+      "num_tokens": 228942799.0,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_math_entropy": 0.1799713587164879,
+      "eval_math_loss": 1.332999587059021,
+      "eval_math_mean_token_accuracy": 0.8159932594299316,
+      "eval_math_num_tokens": 228942799.0,
+      "eval_math_runtime": 35.8043,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8000
+    },
+    {
+      "epoch": 12.8,
+      "eval_physics_entropy": 0.20658580523729325,
+      "eval_physics_loss": 1.3991401195526123,
+      "eval_physics_mean_token_accuracy": 0.7965000495910645,
+      "eval_physics_num_tokens": 228942799.0,
+      "eval_physics_runtime": 41.1662,
+      "eval_physics_samples_per_second": 12.146,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 8000
+    },
+    {
+      "entropy": 0.05222373401047662,
+      "epoch": 12.816,
+      "grad_norm": 0.421875,
+      "learning_rate": 4.424444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902573928236962,
+      "num_tokens": 229230267.0,
+      "step": 8010
+    },
+    {
+      "entropy": 0.05221812457311899,
+      "epoch": 12.832,
+      "grad_norm": 0.45703125,
+      "learning_rate": 4.402222222222223e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902742177248001,
+      "num_tokens": 229513807.0,
+      "step": 8020
+    },
+    {
+      "entropy": 0.05192019739188254,
+      "epoch": 12.848,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.38e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990331218764186,
+      "num_tokens": 229800116.0,
+      "step": 8030
+    },
+    {
+      "entropy": 0.05204707682132721,
+      "epoch": 12.864,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.357777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9904342792928219,
+      "num_tokens": 230086560.0,
+      "step": 8040
+    },
+    {
+      "entropy": 0.05261244847206399,
+      "epoch": 12.88,
+      "grad_norm": 0.4140625,
+      "learning_rate": 4.3355555555555565e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901342626661063,
+      "num_tokens": 230370705.0,
+      "step": 8050
+    },
+    {
+      "entropy": 0.050930758623871954,
+      "epoch": 12.896,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.313333333333334e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904414359480143,
+      "num_tokens": 230665090.0,
+      "step": 8060
+    },
+    {
+      "entropy": 0.0521988584427163,
+      "epoch": 12.912,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.291111111111112e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902524299919605,
+      "num_tokens": 230948774.0,
+      "step": 8070
+    },
+    {
+      "entropy": 0.05130416583269835,
+      "epoch": 12.928,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.268888888888889e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902212977409363,
+      "num_tokens": 231236122.0,
+      "step": 8080
+    },
+    {
+      "entropy": 0.05124916716013104,
+      "epoch": 12.943999999999999,
+      "grad_norm": 0.439453125,
+      "learning_rate": 4.246666666666667e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.990312222391367,
+      "num_tokens": 231520994.0,
+      "step": 8090
+    },
+    {
+      "entropy": 0.05303966203937307,
+      "epoch": 12.96,
+      "grad_norm": 0.474609375,
+      "learning_rate": 4.2244444444444446e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9900147426873446,
+      "num_tokens": 231803509.0,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_math_entropy": 0.18036945784091948,
+      "eval_math_loss": 1.332650899887085,
+      "eval_math_mean_token_accuracy": 0.8161214051246644,
+      "eval_math_num_tokens": 231803509.0,
+      "eval_math_runtime": 35.8002,
+      "eval_math_samples_per_second": 13.966,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8100
+    },
+    {
+      "epoch": 12.96,
+      "eval_physics_entropy": 0.2067739045023918,
+      "eval_physics_loss": 1.3985861539840698,
+      "eval_physics_mean_token_accuracy": 0.7964418997764587,
+      "eval_physics_num_tokens": 231803509.0,
+      "eval_physics_runtime": 41.4682,
+      "eval_physics_samples_per_second": 12.057,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8100
+    },
+    {
+      "entropy": 0.05145827913656831,
+      "epoch": 12.975999999999999,
+      "grad_norm": 0.458984375,
+      "learning_rate": 4.202222222222222e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9905312418937683,
+      "num_tokens": 232095976.0,
+      "step": 8110
+    },
+    {
+      "entropy": 0.05109680832829326,
+      "epoch": 12.992,
+      "grad_norm": 0.453125,
+      "learning_rate": 4.18e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9903466764837503,
+      "num_tokens": 232382972.0,
+      "step": 8120
+    },
+    {
+      "entropy": 0.0512539794202894,
+      "epoch": 13.008,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.157777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906226065009832,
+      "num_tokens": 232672057.0,
+      "step": 8130
+    },
+    {
+      "entropy": 0.050676750240381804,
+      "epoch": 13.024,
+      "grad_norm": 0.388671875,
+      "learning_rate": 4.135555555555556e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.990827102959156,
+      "num_tokens": 232958227.0,
+      "step": 8140
+    },
+    {
+      "entropy": 0.05172369135543704,
+      "epoch": 13.04,
+      "grad_norm": 0.40625,
+      "learning_rate": 4.1133333333333335e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.990977942943573,
+      "num_tokens": 233251931.0,
+      "step": 8150
+    },
+    {
+      "entropy": 0.052580399974249306,
+      "epoch": 13.056,
+      "grad_norm": 0.390625,
+      "learning_rate": 4.091111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905320156365633,
+      "num_tokens": 233531328.0,
+      "step": 8160
+    },
+    {
+      "entropy": 0.05163608979200944,
+      "epoch": 13.072,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.0688888888888896e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990771321207285,
+      "num_tokens": 233816145.0,
+      "step": 8170
+    },
+    {
+      "entropy": 0.05149128795601428,
+      "epoch": 13.088,
+      "grad_norm": 0.404296875,
+      "learning_rate": 4.046666666666667e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.990656491741538,
+      "num_tokens": 234100261.0,
+      "step": 8180
+    },
+    {
+      "entropy": 0.0517811676254496,
+      "epoch": 13.104,
+      "grad_norm": 0.435546875,
+      "learning_rate": 4.024444444444445e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9906980969011784,
+      "num_tokens": 234387753.0,
+      "step": 8190
+    },
+    {
+      "entropy": 0.051599945151247086,
+      "epoch": 13.12,
+      "grad_norm": 0.41015625,
+      "learning_rate": 4.002222222222222e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.990650337934494,
+      "num_tokens": 234673838.0,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_math_entropy": 0.18022768545150758,
+      "eval_math_loss": 1.3355015516281128,
+      "eval_math_mean_token_accuracy": 0.8160842609405518,
+      "eval_math_num_tokens": 234673838.0,
+      "eval_math_runtime": 35.7873,
+      "eval_math_samples_per_second": 13.971,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8200
+    },
+    {
+      "epoch": 13.12,
+      "eval_physics_entropy": 0.20687673270702361,
+      "eval_physics_loss": 1.401163935661316,
+      "eval_physics_mean_token_accuracy": 0.7964475116729737,
+      "eval_physics_num_tokens": 234673838.0,
+      "eval_physics_runtime": 41.4759,
+      "eval_physics_samples_per_second": 12.055,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8200
+    },
+    {
+      "entropy": 0.05104603223735467,
+      "epoch": 13.136,
+      "grad_norm": 0.42578125,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908260378986598,
+      "num_tokens": 234966723.0,
+      "step": 8210
+    },
+    {
+      "entropy": 0.051453781058080496,
+      "epoch": 13.152,
+      "grad_norm": 0.392578125,
+      "learning_rate": 3.9577777777777785e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907778527587652,
+      "num_tokens": 235250894.0,
+      "step": 8220
+    },
+    {
+      "entropy": 0.052391895826440306,
+      "epoch": 13.168,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.935555555555556e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904574226588011,
+      "num_tokens": 235538789.0,
+      "step": 8230
+    },
+    {
+      "entropy": 0.05026549928588793,
+      "epoch": 13.184,
+      "grad_norm": 0.40625,
+      "learning_rate": 3.913333333333334e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909756503999233,
+      "num_tokens": 235837152.0,
+      "step": 8240
+    },
+    {
+      "entropy": 0.053065391920972614,
+      "epoch": 13.2,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.891111111111111e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901943173259496,
+      "num_tokens": 236116206.0,
+      "step": 8250
+    },
+    {
+      "entropy": 0.051446165784727785,
+      "epoch": 13.216,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.868888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906851042062044,
+      "num_tokens": 236399981.0,
+      "step": 8260
+    },
+    {
+      "entropy": 0.0512137760524638,
+      "epoch": 13.232,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.8466666666666665e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906529590487481,
+      "num_tokens": 236687282.0,
+      "step": 8270
+    },
+    {
+      "entropy": 0.05217738675419241,
+      "epoch": 13.248,
+      "grad_norm": 0.5,
+      "learning_rate": 3.824444444444444e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905663672834635,
+      "num_tokens": 236970703.0,
+      "step": 8280
+    },
+    {
+      "entropy": 0.051728677074424925,
+      "epoch": 13.264,
+      "grad_norm": 0.52734375,
+      "learning_rate": 3.8022222222222226e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905138719826937,
+      "num_tokens": 237250995.0,
+      "step": 8290
+    },
+    {
+      "entropy": 0.05051842969842255,
+      "epoch": 13.28,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908084899187088,
+      "num_tokens": 237538556.0,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_math_entropy": 0.1792960543036461,
+      "eval_math_loss": 1.336340308189392,
+      "eval_math_mean_token_accuracy": 0.8159542636871338,
+      "eval_math_num_tokens": 237538556.0,
+      "eval_math_runtime": 35.8026,
+      "eval_math_samples_per_second": 13.965,
+      "eval_math_steps_per_second": 3.491,
+      "step": 8300
+    },
+    {
+      "epoch": 13.28,
+      "eval_physics_entropy": 0.20614614540338516,
+      "eval_physics_loss": 1.4020493030548096,
+      "eval_physics_mean_token_accuracy": 0.796203773021698,
+      "eval_physics_num_tokens": 237538556.0,
+      "eval_physics_runtime": 41.473,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8300
+    },
+    {
+      "entropy": 0.05342790780123323,
+      "epoch": 13.296,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.757777777777778e-06,
+      "loss": 0.0326,
+      "mean_token_accuracy": 0.9900773618370294,
+      "num_tokens": 237813501.0,
+      "step": 8310
+    },
+    {
+      "entropy": 0.05309537169523537,
+      "epoch": 13.312,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.7355555555555555e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903508022427558,
+      "num_tokens": 238094305.0,
+      "step": 8320
+    },
+    {
+      "entropy": 0.05170744662173092,
+      "epoch": 13.328,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.713333333333334e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905386842787266,
+      "num_tokens": 238374175.0,
+      "step": 8330
+    },
+    {
+      "entropy": 0.04992100042290985,
+      "epoch": 13.344,
+      "grad_norm": 0.41015625,
+      "learning_rate": 3.6911111111111115e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908522251993418,
+      "num_tokens": 238668031.0,
+      "step": 8340
+    },
+    {
+      "entropy": 0.05189812278840691,
+      "epoch": 13.36,
+      "grad_norm": 0.470703125,
+      "learning_rate": 3.668888888888889e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9905831549316645,
+      "num_tokens": 238950369.0,
+      "step": 8350
+    },
+    {
+      "entropy": 0.05139565443387255,
+      "epoch": 13.376,
+      "grad_norm": 0.46875,
+      "learning_rate": 3.6466666666666668e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9906327720731497,
+      "num_tokens": 239233027.0,
+      "step": 8360
+    },
+    {
+      "entropy": 0.05204238067381084,
+      "epoch": 13.392,
+      "grad_norm": 0.4609375,
+      "learning_rate": 3.624444444444445e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904446199536323,
+      "num_tokens": 239513675.0,
+      "step": 8370
+    },
+    {
+      "entropy": 0.051922239165287466,
+      "epoch": 13.408,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.6022222222222224e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9902938231825829,
+      "num_tokens": 239797199.0,
+      "step": 8380
+    },
+    {
+      "entropy": 0.05086992103606462,
+      "epoch": 13.424,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.58e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905360575765372,
+      "num_tokens": 240082617.0,
+      "step": 8390
+    },
+    {
+      "entropy": 0.052436576853506264,
+      "epoch": 13.44,
+      "grad_norm": 0.455078125,
+      "learning_rate": 3.5577777777777785e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9903819940984249,
+      "num_tokens": 240365037.0,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_math_entropy": 0.17929287600517274,
+      "eval_math_loss": 1.3361586332321167,
+      "eval_math_mean_token_accuracy": 0.8159851579666137,
+      "eval_math_num_tokens": 240365037.0,
+      "eval_math_runtime": 35.8812,
+      "eval_math_samples_per_second": 13.935,
+      "eval_math_steps_per_second": 3.484,
+      "step": 8400
+    },
+    {
+      "epoch": 13.44,
+      "eval_physics_entropy": 0.2061038536429405,
+      "eval_physics_loss": 1.4018964767456055,
+      "eval_physics_mean_token_accuracy": 0.7964254455566406,
+      "eval_physics_num_tokens": 240365037.0,
+      "eval_physics_runtime": 41.5825,
+      "eval_physics_samples_per_second": 12.024,
+      "eval_physics_steps_per_second": 3.006,
+      "step": 8400
+    },
+    {
+      "entropy": 0.052103328774683176,
+      "epoch": 13.456,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.535555555555556e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9904427032917738,
+      "num_tokens": 240647742.0,
+      "step": 8410
+    },
+    {
+      "entropy": 0.0523280065273866,
+      "epoch": 13.472,
+      "grad_norm": 0.44921875,
+      "learning_rate": 3.5133333333333337e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9904405385255813,
+      "num_tokens": 240926943.0,
+      "step": 8420
+    },
+    {
+      "entropy": 0.05302672478137538,
+      "epoch": 13.488,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4911111111111113e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9903274435549975,
+      "num_tokens": 241206172.0,
+      "step": 8430
+    },
+    {
+      "entropy": 0.05202294115442783,
+      "epoch": 13.504,
+      "grad_norm": 0.48046875,
+      "learning_rate": 3.4688888888888894e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9905512318015098,
+      "num_tokens": 241489988.0,
+      "step": 8440
+    },
+    {
+      "entropy": 0.050564923824276777,
+      "epoch": 13.52,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.446666666666667e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907515056431293,
+      "num_tokens": 241781801.0,
+      "step": 8450
+    },
+    {
+      "entropy": 0.05287049149628729,
+      "epoch": 13.536,
+      "grad_norm": 0.50390625,
+      "learning_rate": 3.4244444444444446e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9903701849281787,
+      "num_tokens": 242064444.0,
+      "step": 8460
+    },
+    {
+      "entropy": 0.05167013480095193,
+      "epoch": 13.552,
+      "grad_norm": 0.4765625,
+      "learning_rate": 3.4022222222222222e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9904057901352644,
+      "num_tokens": 242354272.0,
+      "step": 8470
+    },
+    {
+      "entropy": 0.0516669807722792,
+      "epoch": 13.568,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9903885431587696,
+      "num_tokens": 242640475.0,
+      "step": 8480
+    },
+    {
+      "entropy": 0.05292565480340272,
+      "epoch": 13.584,
+      "grad_norm": 0.4375,
+      "learning_rate": 3.3577777777777783e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9901615660637617,
+      "num_tokens": 242920673.0,
+      "step": 8490
+    },
+    {
+      "entropy": 0.051914384588599205,
+      "epoch": 13.6,
+      "grad_norm": 0.416015625,
+      "learning_rate": 3.335555555555556e-06,
+      "loss": 0.0322,
+      "mean_token_accuracy": 0.990172453969717,
+      "num_tokens": 243198176.0,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_math_entropy": 0.17909148234128952,
+      "eval_math_loss": 1.3365639448165894,
+      "eval_math_mean_token_accuracy": 0.8161025867462158,
+      "eval_math_num_tokens": 243198176.0,
+      "eval_math_runtime": 35.7973,
+      "eval_math_samples_per_second": 13.968,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8500
+    },
+    {
+      "epoch": 13.6,
+      "eval_physics_entropy": 0.2059074219465256,
+      "eval_physics_loss": 1.4022636413574219,
+      "eval_physics_mean_token_accuracy": 0.7963426208496094,
+      "eval_physics_num_tokens": 243198176.0,
+      "eval_physics_runtime": 41.1599,
+      "eval_physics_samples_per_second": 12.148,
+      "eval_physics_steps_per_second": 3.037,
+      "step": 8500
+    },
+    {
+      "entropy": 0.05167351650306955,
+      "epoch": 13.616,
+      "grad_norm": 0.478515625,
+      "learning_rate": 3.3133333333333335e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904801592230796,
+      "num_tokens": 243483881.0,
+      "step": 8510
+    },
+    {
+      "entropy": 0.053366528800688685,
+      "epoch": 13.632,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.2911111111111116e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9902401711791754,
+      "num_tokens": 243764128.0,
+      "step": 8520
+    },
+    {
+      "entropy": 0.05087098733056337,
+      "epoch": 13.648,
+      "grad_norm": 0.443359375,
+      "learning_rate": 3.268888888888889e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904976122081279,
+      "num_tokens": 244051585.0,
+      "step": 8530
+    },
+    {
+      "entropy": 0.05087883236119524,
+      "epoch": 13.664,
+      "grad_norm": 0.447265625,
+      "learning_rate": 3.2466666666666668e-06,
+      "loss": 0.0302,
+      "mean_token_accuracy": 0.9905655104666948,
+      "num_tokens": 244349004.0,
+      "step": 8540
+    },
+    {
+      "entropy": 0.04904007929144427,
+      "epoch": 13.68,
+      "grad_norm": 0.4453125,
+      "learning_rate": 3.2244444444444444e-06,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.990977830812335,
+      "num_tokens": 244649637.0,
+      "step": 8550
+    },
+    {
+      "entropy": 0.0504306229762733,
+      "epoch": 13.696,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.202222222222223e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907431546598673,
+      "num_tokens": 244940362.0,
+      "step": 8560
+    },
+    {
+      "entropy": 0.05084956689970568,
+      "epoch": 13.712,
+      "grad_norm": 0.421875,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904598340392112,
+      "num_tokens": 245228064.0,
+      "step": 8570
+    },
+    {
+      "entropy": 0.05026600864948705,
+      "epoch": 13.728,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.157777777777778e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9905784897506237,
+      "num_tokens": 245516918.0,
+      "step": 8580
+    },
+    {
+      "entropy": 0.050801993359345946,
+      "epoch": 13.744,
+      "grad_norm": 0.458984375,
+      "learning_rate": 3.1355555555555557e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907515291124582,
+      "num_tokens": 245807811.0,
+      "step": 8590
+    },
+    {
+      "entropy": 0.05082948401104659,
+      "epoch": 13.76,
+      "grad_norm": 0.423828125,
+      "learning_rate": 3.1133333333333337e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9906744692474604,
+      "num_tokens": 246099779.0,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_math_entropy": 0.17962482076883315,
+      "eval_math_loss": 1.3373396396636963,
+      "eval_math_mean_token_accuracy": 0.816063407421112,
+      "eval_math_num_tokens": 246099779.0,
+      "eval_math_runtime": 35.7911,
+      "eval_math_samples_per_second": 13.97,
+      "eval_math_steps_per_second": 3.492,
+      "step": 8600
+    },
+    {
+      "epoch": 13.76,
+      "eval_physics_entropy": 0.20641914105415343,
+      "eval_physics_loss": 1.4027777910232544,
+      "eval_physics_mean_token_accuracy": 0.796207398891449,
+      "eval_physics_num_tokens": 246099779.0,
+      "eval_physics_runtime": 41.482,
+      "eval_physics_samples_per_second": 12.053,
+      "eval_physics_steps_per_second": 3.013,
+      "step": 8600
+    },
+    {
+      "entropy": 0.05176880229264498,
+      "epoch": 13.776,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0911111111111114e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904642947018146,
+      "num_tokens": 246383665.0,
+      "step": 8610
+    },
+    {
+      "entropy": 0.05162258397322148,
+      "epoch": 13.792,
+      "grad_norm": 0.44140625,
+      "learning_rate": 3.068888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904142130166292,
+      "num_tokens": 246668395.0,
+      "step": 8620
+    },
+    {
+      "entropy": 0.0508142440347001,
+      "epoch": 13.808,
+      "grad_norm": 0.40234375,
+      "learning_rate": 3.0466666666666666e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9907273758202791,
+      "num_tokens": 246960911.0,
+      "step": 8630
+    },
+    {
+      "entropy": 0.04985311826458201,
+      "epoch": 13.824,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.024444444444445e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908174972981214,
+      "num_tokens": 247253562.0,
+      "step": 8640
+    },
+    {
+      "entropy": 0.052733641443774106,
+      "epoch": 13.84,
+      "grad_norm": 0.462890625,
+      "learning_rate": 3.0022222222222227e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9900983639061451,
+      "num_tokens": 247529533.0,
+      "step": 8650
+    },
+    {
+      "entropy": 0.05027402397245169,
+      "epoch": 13.856,
+      "grad_norm": 0.45703125,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.990868479758501,
+      "num_tokens": 247819702.0,
+      "step": 8660
+    },
+    {
+      "entropy": 0.050137166888453065,
+      "epoch": 13.872,
+      "grad_norm": 0.451171875,
+      "learning_rate": 2.957777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907190252095461,
+      "num_tokens": 248113990.0,
+      "step": 8670
+    },
+    {
+      "entropy": 0.0517217613523826,
+      "epoch": 13.888,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.935555555555556e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9902944087982177,
+      "num_tokens": 248397281.0,
+      "step": 8680
+    },
+    {
+      "entropy": 0.05074853647965938,
+      "epoch": 13.904,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.9133333333333335e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9904990620911122,
+      "num_tokens": 248689496.0,
+      "step": 8690
+    },
+    {
+      "entropy": 0.05088953715749085,
+      "epoch": 13.92,
+      "grad_norm": 0.431640625,
+      "learning_rate": 2.891111111111111e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905771657824516,
+      "num_tokens": 248975741.0,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_math_entropy": 0.17970398700237275,
+      "eval_math_loss": 1.336698293685913,
+      "eval_math_mean_token_accuracy": 0.8159835457801818,
+      "eval_math_num_tokens": 248975741.0,
+      "eval_math_runtime": 35.7809,
+      "eval_math_samples_per_second": 13.974,
+      "eval_math_steps_per_second": 3.493,
+      "step": 8700
+    },
+    {
+      "epoch": 13.92,
+      "eval_physics_entropy": 0.20640488415956498,
+      "eval_physics_loss": 1.4024730920791626,
+      "eval_physics_mean_token_accuracy": 0.7963580441474915,
+      "eval_physics_num_tokens": 248975741.0,
+      "eval_physics_runtime": 41.4599,
+      "eval_physics_samples_per_second": 12.06,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8700
+    },
+    {
+      "entropy": 0.04994275859789923,
+      "epoch": 13.936,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.868888888888889e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9907119512557984,
+      "num_tokens": 249270221.0,
+      "step": 8710
+    },
+    {
+      "entropy": 0.04963390504708513,
+      "epoch": 13.952,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.8466666666666672e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9907703123986721,
+      "num_tokens": 249563652.0,
+      "step": 8720
+    },
+    {
+      "entropy": 0.05193966852966696,
+      "epoch": 13.968,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.824444444444445e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9902710720896721,
+      "num_tokens": 249847652.0,
+      "step": 8730
+    },
+    {
+      "entropy": 0.05370221418561414,
+      "epoch": 13.984,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.8022222222222225e-06,
+      "loss": 0.033,
+      "mean_token_accuracy": 0.9897177372127771,
+      "num_tokens": 250120462.0,
+      "step": 8740
+    },
+    {
+      "entropy": 0.05114520986098796,
+      "epoch": 14.0,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9904374986886978,
+      "num_tokens": 250409936.0,
+      "step": 8750
+    },
+    {
+      "entropy": 0.05091056008823216,
+      "epoch": 14.016,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.757777777777778e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908093482255935,
+      "num_tokens": 250696959.0,
+      "step": 8760
+    },
+    {
+      "entropy": 0.052239940082654354,
+      "epoch": 14.032,
+      "grad_norm": 0.4296875,
+      "learning_rate": 2.7355555555555557e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906717732548713,
+      "num_tokens": 250980823.0,
+      "step": 8770
+    },
+    {
+      "entropy": 0.051255598349962385,
+      "epoch": 14.048,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.7133333333333333e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.990817604586482,
+      "num_tokens": 251268212.0,
+      "step": 8780
+    },
+    {
+      "entropy": 0.05210095479851588,
+      "epoch": 14.064,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.6911111111111114e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9904542353004218,
+      "num_tokens": 251553952.0,
+      "step": 8790
+    },
+    {
+      "entropy": 0.05244319257326424,
+      "epoch": 14.08,
+      "grad_norm": 0.458984375,
+      "learning_rate": 2.6688888888888894e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9904410917311907,
+      "num_tokens": 251838168.0,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_math_entropy": 0.17896863031387328,
+      "eval_math_loss": 1.3373428583145142,
+      "eval_math_mean_token_accuracy": 0.8160420455932618,
+      "eval_math_num_tokens": 251838168.0,
+      "eval_math_runtime": 35.776,
+      "eval_math_samples_per_second": 13.976,
+      "eval_math_steps_per_second": 3.494,
+      "step": 8800
+    },
+    {
+      "epoch": 14.08,
+      "eval_physics_entropy": 0.20583795410394667,
+      "eval_physics_loss": 1.4028042554855347,
+      "eval_physics_mean_token_accuracy": 0.7963007631301879,
+      "eval_physics_num_tokens": 251838168.0,
+      "eval_physics_runtime": 41.4723,
+      "eval_physics_samples_per_second": 12.056,
+      "eval_physics_steps_per_second": 3.014,
+      "step": 8800
+    },
+    {
+      "entropy": 0.05141720774117857,
+      "epoch": 14.096,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.646666666666667e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906542908400298,
+      "num_tokens": 252121559.0,
+      "step": 8810
+    },
+    {
+      "entropy": 0.050573783030267806,
+      "epoch": 14.112,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.6244444444444446e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9908661898225546,
+      "num_tokens": 252413324.0,
+      "step": 8820
+    },
+    {
+      "entropy": 0.05072221851442009,
+      "epoch": 14.128,
+      "grad_norm": 0.50390625,
+      "learning_rate": 2.6022222222222227e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9908107925206423,
+      "num_tokens": 252700652.0,
+      "step": 8830
+    },
+    {
+      "entropy": 0.04975284937536344,
+      "epoch": 14.144,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910169199109078,
+      "num_tokens": 252992593.0,
+      "step": 8840
+    },
+    {
+      "entropy": 0.049413899821229276,
+      "epoch": 14.16,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.557777777777778e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910381846129894,
+      "num_tokens": 253284586.0,
+      "step": 8850
+    },
+    {
+      "entropy": 0.04987312061712146,
+      "epoch": 14.176,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.5355555555555555e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.9910611238330602,
+      "num_tokens": 253580393.0,
+      "step": 8860
+    },
+    {
+      "entropy": 0.05002628688234836,
+      "epoch": 14.192,
+      "grad_norm": 0.416015625,
+      "learning_rate": 2.5133333333333336e-06,
+      "loss": 0.0303,
+      "mean_token_accuracy": 0.9909944098442793,
+      "num_tokens": 253873719.0,
+      "step": 8870
+    },
+    {
+      "entropy": 0.051363390649203214,
+      "epoch": 14.208,
+      "grad_norm": 0.42578125,
+      "learning_rate": 2.491111111111111e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.990709126740694,
+      "num_tokens": 254163170.0,
+      "step": 8880
+    },
+    {
+      "entropy": 0.05138509668176994,
+      "epoch": 14.224,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.468888888888889e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905464217066765,
+      "num_tokens": 254445476.0,
+      "step": 8890
+    },
+    {
+      "entropy": 0.051887833909131584,
+      "epoch": 14.24,
+      "grad_norm": 0.455078125,
+      "learning_rate": 2.446666666666667e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9906827192753553,
+      "num_tokens": 254731054.0,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_math_entropy": 0.17892871648073197,
+      "eval_math_loss": 1.3377963304519653,
+      "eval_math_mean_token_accuracy": 0.8161516799926758,
+      "eval_math_num_tokens": 254731054.0,
+      "eval_math_runtime": 35.7685,
+      "eval_math_samples_per_second": 13.979,
+      "eval_math_steps_per_second": 3.495,
+      "step": 8900
+    },
+    {
+      "epoch": 14.24,
+      "eval_physics_entropy": 0.20582070446014405,
+      "eval_physics_loss": 1.4032886028289795,
+      "eval_physics_mean_token_accuracy": 0.7961821508407593,
+      "eval_physics_num_tokens": 254731054.0,
+      "eval_physics_runtime": 41.4536,
+      "eval_physics_samples_per_second": 12.062,
+      "eval_physics_steps_per_second": 3.015,
+      "step": 8900
+    },
+    {
+      "entropy": 0.05053608596790582,
+      "epoch": 14.256,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.4244444444444444e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908701453357935,
+      "num_tokens": 255017263.0,
+      "step": 8910
+    },
+    {
+      "entropy": 0.04994903282495215,
+      "epoch": 14.272,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.4022222222222225e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907931432127952,
+      "num_tokens": 255310045.0,
+      "step": 8920
+    },
+    {
+      "entropy": 0.0498261819826439,
+      "epoch": 14.288,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.38e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908365543931723,
+      "num_tokens": 255601380.0,
+      "step": 8930
+    },
+    {
+      "entropy": 0.05181859011063352,
+      "epoch": 14.304,
+      "grad_norm": 0.474609375,
+      "learning_rate": 2.357777777777778e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906132340431213,
+      "num_tokens": 255886699.0,
+      "step": 8940
+    },
+    {
+      "entropy": 0.0507721847621724,
+      "epoch": 14.32,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.3355555555555557e-06,
+      "loss": 0.0307,
+      "mean_token_accuracy": 0.9908549074083567,
+      "num_tokens": 256174927.0,
+      "step": 8950
+    },
+    {
+      "entropy": 0.05199292111210525,
+      "epoch": 14.336,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.3133333333333333e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.990635683387518,
+      "num_tokens": 256456044.0,
+      "step": 8960
+    },
+    {
+      "entropy": 0.051074851502198725,
+      "epoch": 14.352,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.2911111111111114e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9907376989722252,
+      "num_tokens": 256737497.0,
+      "step": 8970
+    },
+    {
+      "entropy": 0.050048362242523584,
+      "epoch": 14.368,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.268888888888889e-06,
+      "loss": 0.03,
+      "mean_token_accuracy": 0.9909071121364832,
+      "num_tokens": 257032721.0,
+      "step": 8980
+    },
+    {
+      "entropy": 0.052360807289369406,
+      "epoch": 14.384,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.2466666666666666e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.9905145078897476,
+      "num_tokens": 257311728.0,
+      "step": 8990
+    },
+    {
+      "entropy": 0.051360888360068205,
+      "epoch": 14.4,
+      "grad_norm": 0.4921875,
+      "learning_rate": 2.2244444444444447e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906007140874863,
+      "num_tokens": 257598724.0,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_math_entropy": 0.17880775159597398,
+      "eval_math_loss": 1.3379243612289429,
+      "eval_math_mean_token_accuracy": 0.8160511145591736,
+      "eval_math_num_tokens": 257598724.0,
+      "eval_math_runtime": 35.7772,
+      "eval_math_samples_per_second": 13.975,
+      "eval_math_steps_per_second": 3.494,
+      "step": 9000
+    },
+    {
+      "epoch": 14.4,
+      "eval_physics_entropy": 0.20577423387765884,
+      "eval_physics_loss": 1.4032671451568604,
+      "eval_physics_mean_token_accuracy": 0.7963672475814819,
+      "eval_physics_num_tokens": 257598724.0,
+      "eval_physics_runtime": 41.5292,
+      "eval_physics_samples_per_second": 12.04,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 9000
+    },
+    {
+      "entropy": 0.05165425557643175,
+      "epoch": 14.416,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.2022222222222227e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9904982160776854,
+      "num_tokens": 257881999.0,
+      "step": 9010
+    },
+    {
+      "entropy": 0.05141539016040042,
+      "epoch": 14.432,
+      "grad_norm": 0.439453125,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9906045537441969,
+      "num_tokens": 258168215.0,
+      "step": 9020
+    },
+    {
+      "entropy": 0.050296508916653694,
+      "epoch": 14.448,
+      "grad_norm": 0.375,
+      "learning_rate": 2.157777777777778e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907093074172735,
+      "num_tokens": 258455416.0,
+      "step": 9030
+    },
+    {
+      "entropy": 0.050917042326182126,
+      "epoch": 14.464,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.1355555555555555e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907846868038177,
+      "num_tokens": 258743229.0,
+      "step": 9040
+    },
+    {
+      "entropy": 0.050571117317304015,
+      "epoch": 14.48,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.1133333333333336e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9907037664204836,
+      "num_tokens": 259028366.0,
+      "step": 9050
+    },
+    {
+      "entropy": 0.05255825007334351,
+      "epoch": 14.496,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.091111111111111e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.990223740413785,
+      "num_tokens": 259304539.0,
+      "step": 9060
+    },
+    {
+      "entropy": 0.04984008674509823,
+      "epoch": 14.512,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.0688888888888892e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9909922797232866,
+      "num_tokens": 259596545.0,
+      "step": 9070
+    },
+    {
+      "entropy": 0.05399213619530201,
+      "epoch": 14.528,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.046666666666667e-06,
+      "loss": 0.0323,
+      "mean_token_accuracy": 0.9901899788528681,
+      "num_tokens": 259870582.0,
+      "step": 9080
+    },
+    {
+      "entropy": 0.05196048454381526,
+      "epoch": 14.544,
+      "grad_norm": 0.47265625,
+      "learning_rate": 2.024444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904056012630462,
+      "num_tokens": 260152920.0,
+      "step": 9090
+    },
+    {
+      "entropy": 0.05186033591162413,
+      "epoch": 14.56,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.0022222222222225e-06,
+      "loss": 0.0313,
+      "mean_token_accuracy": 0.9905133057385683,
+      "num_tokens": 260437753.0,
+      "step": 9100
+    },
+    {
+      "epoch": 14.56,
+      "eval_math_entropy": 0.1789300767183304,
+      "eval_math_loss": 1.337916612625122,
+      "eval_math_mean_token_accuracy": 0.8160979781150818,
+      "eval_math_num_tokens": 260437753.0,
+      "eval_math_runtime": 35.8923,
+      "eval_math_samples_per_second": 13.931,
+      "eval_math_steps_per_second": 3.483,
+      "step": 9100
+    },
+    {
+      "epoch": 14.56,
+      "eval_physics_entropy": 0.2058382231593132,
+      "eval_physics_loss": 1.4032976627349854,
+      "eval_physics_mean_token_accuracy": 0.796412097454071,
+      "eval_physics_num_tokens": 260437753.0,
+      "eval_physics_runtime": 41.6691,
+      "eval_physics_samples_per_second": 11.999,
+      "eval_physics_steps_per_second": 3.0,
+      "step": 9100
+    },
+    {
+      "entropy": 0.05178943979553878,
+      "epoch": 14.576,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.98e-06,
+      "loss": 0.0315,
+      "mean_token_accuracy": 0.9905171319842339,
+      "num_tokens": 260719119.0,
+      "step": 9110
+    },
+    {
+      "entropy": 0.05121098265517503,
+      "epoch": 14.592,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.9577777777777777e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9907041292637586,
+      "num_tokens": 261005643.0,
+      "step": 9120
+    },
+    {
+      "entropy": 0.05071252475026995,
+      "epoch": 14.608,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.9355555555555558e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908163469284773,
+      "num_tokens": 261297751.0,
+      "step": 9130
+    },
+    {
+      "entropy": 0.052167282067239286,
+      "epoch": 14.624,
+      "grad_norm": 0.5078125,
+      "learning_rate": 1.9133333333333334e-06,
+      "loss": 0.032,
+      "mean_token_accuracy": 0.9905383501201868,
+      "num_tokens": 261574524.0,
+      "step": 9140
+    },
+    {
+      "entropy": 0.05092277757357806,
+      "epoch": 14.64,
+      "grad_norm": 0.4609375,
+      "learning_rate": 1.8911111111111114e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906485483050347,
+      "num_tokens": 261860827.0,
+      "step": 9150
+    },
+    {
+      "entropy": 0.0508843480842188,
+      "epoch": 14.656,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.868888888888889e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9905889291316271,
+      "num_tokens": 262146577.0,
+      "step": 9160
+    },
+    {
+      "entropy": 0.051762618916109206,
+      "epoch": 14.672,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.8466666666666668e-06,
+      "loss": 0.0319,
+      "mean_token_accuracy": 0.990484931319952,
+      "num_tokens": 262425056.0,
+      "step": 9170
+    },
+    {
+      "entropy": 0.050335659622214736,
+      "epoch": 14.688,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.8244444444444445e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9908275879919529,
+      "num_tokens": 262718069.0,
+      "step": 9180
+    },
+    {
+      "entropy": 0.05197438263567165,
+      "epoch": 14.704,
+      "grad_norm": 0.404296875,
+      "learning_rate": 1.8022222222222225e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.990571141988039,
+      "num_tokens": 263005354.0,
+      "step": 9190
+    },
+    {
+      "entropy": 0.050634497124701736,
+      "epoch": 14.72,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9907538581639528,
+      "num_tokens": 263290156.0,
+      "step": 9200
+    },
+    {
+      "epoch": 14.72,
+      "eval_math_entropy": 0.17962117153406143,
+      "eval_math_loss": 1.3382362127304077,
+      "eval_math_mean_token_accuracy": 0.8159899268150329,
+      "eval_math_num_tokens": 263290156.0,
+      "eval_math_runtime": 35.8309,
+      "eval_math_samples_per_second": 13.954,
+      "eval_math_steps_per_second": 3.489,
+      "step": 9200
+    },
+    {
+      "epoch": 14.72,
+      "eval_physics_entropy": 0.20646389347314834,
+      "eval_physics_loss": 1.4038169384002686,
+      "eval_physics_mean_token_accuracy": 0.7962654252052307,
+      "eval_physics_num_tokens": 263290156.0,
+      "eval_physics_runtime": 41.5336,
+      "eval_physics_samples_per_second": 12.038,
+      "eval_physics_steps_per_second": 3.01,
+      "step": 9200
+    },
+    {
+      "entropy": 0.05221541177015752,
+      "epoch": 14.736,
+      "grad_norm": 0.4296875,
+      "learning_rate": 1.757777777777778e-06,
+      "loss": 0.0316,
+      "mean_token_accuracy": 0.9905163571238518,
+      "num_tokens": 263570163.0,
+      "step": 9210
+    },
+    {
+      "entropy": 0.05265975071815774,
+      "epoch": 14.752,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.7355555555555555e-06,
+      "loss": 0.0321,
+      "mean_token_accuracy": 0.9901549957692624,
+      "num_tokens": 263847082.0,
+      "step": 9220
+    },
+    {
+      "entropy": 0.05067356282379478,
+      "epoch": 14.768,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.7133333333333336e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9906387656927109,
+      "num_tokens": 264135491.0,
+      "step": 9230
+    },
+    {
+      "entropy": 0.05214657548349351,
+      "epoch": 14.784,
+      "grad_norm": 0.51171875,
+      "learning_rate": 1.6911111111111112e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9905820898711681,
+      "num_tokens": 264423201.0,
+      "step": 9240
+    },
+    {
+      "entropy": 0.05079062914010137,
+      "epoch": 14.8,
+      "grad_norm": 0.431640625,
+      "learning_rate": 1.668888888888889e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907969791442156,
+      "num_tokens": 264711872.0,
+      "step": 9250
+    },
+    {
+      "entropy": 0.05266349882585928,
+      "epoch": 14.816,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.6466666666666666e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9903959967195988,
+      "num_tokens": 264991504.0,
+      "step": 9260
+    },
+    {
+      "entropy": 0.05068820312153548,
+      "epoch": 14.832,
+      "grad_norm": 0.375,
+      "learning_rate": 1.6244444444444447e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9907281890511512,
+      "num_tokens": 265283873.0,
+      "step": 9270
+    },
+    {
+      "entropy": 0.05143596041016281,
+      "epoch": 14.848,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.6022222222222223e-06,
+      "loss": 0.0318,
+      "mean_token_accuracy": 0.9905165329575538,
+      "num_tokens": 265563990.0,
+      "step": 9280
+    },
+    {
+      "entropy": 0.05017106860177591,
+      "epoch": 14.864,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906800303608179,
+      "num_tokens": 265850179.0,
+      "step": 9290
+    },
+    {
+      "entropy": 0.04978077416308224,
+      "epoch": 14.88,
+      "grad_norm": 0.470703125,
+      "learning_rate": 1.5577777777777777e-06,
+      "loss": 0.0301,
+      "mean_token_accuracy": 0.9907668430358172,
+      "num_tokens": 266145551.0,
+      "step": 9300
+    },
+    {
+      "epoch": 14.88,
+      "eval_math_entropy": 0.17940663027763368,
+      "eval_math_loss": 1.3379700183868408,
+      "eval_math_mean_token_accuracy": 0.8160032463073731,
+      "eval_math_num_tokens": 266145551.0,
+      "eval_math_runtime": 35.8577,
+      "eval_math_samples_per_second": 13.944,
+      "eval_math_steps_per_second": 3.486,
+      "step": 9300
+    },
+    {
+      "epoch": 14.88,
+      "eval_physics_entropy": 0.20622114306688308,
+      "eval_physics_loss": 1.4035049676895142,
+      "eval_physics_mean_token_accuracy": 0.7961468186378479,
+      "eval_physics_num_tokens": 266145551.0,
+      "eval_physics_runtime": 41.5608,
+      "eval_physics_samples_per_second": 12.031,
+      "eval_physics_steps_per_second": 3.008,
+      "step": 9300
+    },
+    {
+      "entropy": 0.050715955626219514,
+      "epoch": 14.896,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.5355555555555558e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9907834134995938,
+      "num_tokens": 266432255.0,
+      "step": 9310
+    },
+    {
+      "entropy": 0.0511104385368526,
+      "epoch": 14.912,
+      "grad_norm": 0.50390625,
+      "learning_rate": 1.5133333333333334e-06,
+      "loss": 0.031,
+      "mean_token_accuracy": 0.9906729698181153,
+      "num_tokens": 266719617.0,
+      "step": 9320
+    },
+    {
+      "entropy": 0.0505100853391923,
+      "epoch": 14.928,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.4911111111111112e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9907874137163162,
+      "num_tokens": 267012400.0,
+      "step": 9330
+    },
+    {
+      "entropy": 0.05214700745418668,
+      "epoch": 14.943999999999999,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.468888888888889e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904575120657683,
+      "num_tokens": 267295468.0,
+      "step": 9340
+    },
+    {
+      "entropy": 0.051390086626634,
+      "epoch": 14.96,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.4466666666666669e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905224211513997,
+      "num_tokens": 267580054.0,
+      "step": 9350
+    },
+    {
+      "entropy": 0.050726299383677544,
+      "epoch": 14.975999999999999,
+      "grad_norm": 0.462890625,
+      "learning_rate": 1.4244444444444447e-06,
+      "loss": 0.0305,
+      "mean_token_accuracy": 0.9908859673887491,
+      "num_tokens": 267871906.0,
+      "step": 9360
+    },
+    {
+      "entropy": 0.05101369803305715,
+      "epoch": 14.992,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.4022222222222223e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9905830901116133,
+      "num_tokens": 268156343.0,
+      "step": 9370
+    },
+    {
+      "entropy": 0.051476930570788684,
+      "epoch": 15.008,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.0311,
+      "mean_token_accuracy": 0.9904872190207243,
+      "num_tokens": 268442263.0,
+      "step": 9380
+    },
+    {
+      "entropy": 0.05267978028859943,
+      "epoch": 15.024,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.357777777777778e-06,
+      "loss": 0.0324,
+      "mean_token_accuracy": 0.990261047706008,
+      "num_tokens": 268714841.0,
+      "step": 9390
+    },
+    {
+      "entropy": 0.05263522565364838,
+      "epoch": 15.04,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.3355555555555558e-06,
+      "loss": 0.0317,
+      "mean_token_accuracy": 0.9904483895748853,
+      "num_tokens": 268994210.0,
+      "step": 9400
+    },
+    {
+      "epoch": 15.04,
+      "eval_math_entropy": 0.17955057173967362,
+      "eval_math_loss": 1.3379791975021362,
+      "eval_math_mean_token_accuracy": 0.8160371327400208,
+      "eval_math_num_tokens": 268994210.0,
+      "eval_math_runtime": 35.8381,
+      "eval_math_samples_per_second": 13.952,
+      "eval_math_steps_per_second": 3.488,
+      "step": 9400
+    },
+    {
+      "epoch": 15.04,
+      "eval_physics_entropy": 0.20631873172521592,
+      "eval_physics_loss": 1.403602957725525,
+      "eval_physics_mean_token_accuracy": 0.7963659038543701,
+      "eval_physics_num_tokens": 268994210.0,
+      "eval_physics_runtime": 41.5687,
+      "eval_physics_samples_per_second": 12.028,
+      "eval_physics_steps_per_second": 3.007,
+      "step": 9400
+    },
+    {
+      "entropy": 0.05078244159230962,
+      "epoch": 15.056,
+      "grad_norm": 0.376953125,
+      "learning_rate": 1.3133333333333334e-06,
+      "loss": 0.0304,
+      "mean_token_accuracy": 0.9910456586629153,
+      "num_tokens": 269285822.0,
+      "step": 9410
+    },
+    {
+      "entropy": 0.04946443063672632,
+      "epoch": 15.072,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.2911111111111112e-06,
+      "loss": 0.0303,
+      "mean_token_accuracy": 0.9911473792046308,
+      "num_tokens": 269579649.0,
+      "step": 9420
+    },
+    {
+      "entropy": 0.050997486570850015,
+      "epoch": 15.088,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.268888888888889e-06,
+      "loss": 0.0308,
+      "mean_token_accuracy": 0.9908084504306316,
+      "num_tokens": 269866946.0,
+      "step": 9430
+    },
+    {
+      "entropy": 0.051223152736201885,
+      "epoch": 15.104,
+      "grad_norm": 0.484375,
+      "learning_rate": 1.2466666666666667e-06,
+      "loss": 0.0312,
+      "mean_token_accuracy": 0.9909345023334026,
+      "num_tokens": 270151076.0,
+      "step": 9440
+    },
+    {
+      "entropy": 0.05211627440294251,
+      "epoch": 15.12,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.2244444444444445e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9906228743493557,
+      "num_tokens": 270433447.0,
+      "step": 9450
+    },
+    {
+      "entropy": 0.05138029637746513,
+      "epoch": 15.136,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.2022222222222223e-06,
+      "loss": 0.0309,
+      "mean_token_accuracy": 0.9908742789179087,
+      "num_tokens": 270721393.0,
+      "step": 9460
+    },
+    {
+      "entropy": 0.05007477108156309,
+      "epoch": 15.152,
+      "grad_norm": 0.478515625,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 0.0298,
+      "mean_token_accuracy": 0.9911369781941175,
+      "num_tokens": 271018581.0,
+      "step": 9470
+    },
+    {
+      "entropy": 0.050468166405335066,
+      "epoch": 15.168,
+      "grad_norm": 0.388671875,
+      "learning_rate": 1.1577777777777778e-06,
+      "loss": 0.0306,
+      "mean_token_accuracy": 0.9908524744212628,
+      "num_tokens": 271307471.0,
+      "step": 9480
+    },
+    {
+      "entropy": 0.05269346619024873,
+      "epoch": 15.184,
+      "grad_norm": 0.447265625,
+      "learning_rate": 1.1355555555555558e-06,
+      "loss": 0.0314,
+      "mean_token_accuracy": 0.9904394499957562,
+      "num_tokens": 271590009.0,
+      "step": 9490
+    },
+    {
+      "entropy": 0.050732248765416446,
+      "epoch": 15.2,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.1133333333333334e-06,
+      "loss": 0.0302,
+      "mean_token_accuracy": 0.9909780859947205,
+      "num_tokens": 271882895.0,
+      "step": 9500
+    },
+    {
+      "epoch": 15.2,
+      "eval_math_entropy": 0.17890966862440108,
+      "eval_math_loss": 1.3383686542510986,
+      "eval_math_mean_token_accuracy": 0.8160392146110534,
+      "eval_math_num_tokens": 271882895.0,
+      "eval_math_runtime": 35.8479,
+      "eval_math_samples_per_second": 13.948,
+      "eval_math_steps_per_second": 3.487,
+      "step": 9500
+    },
+    {
+      "epoch": 15.2,
+      "eval_physics_entropy": 0.205828775703907,
+      "eval_physics_loss": 1.4038267135620117,
+      "eval_physics_mean_token_accuracy": 0.7963502130508423,
+      "eval_physics_num_tokens": 271882895.0,
+      "eval_physics_runtime": 41.1707,
+      "eval_physics_samples_per_second": 12.145,
+      "eval_physics_steps_per_second": 3.036,
+      "step": 9500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 16,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.061325380407013e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/checkpoint-9500/training_args.bin b/checkpoint-9500/training_args.bin
new file mode 100644
index 0000000..8d77757
--- /dev/null
+++ b/checkpoint-9500/training_args.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0612b24a8c5a6d7aa7710b98af44bc59fdb9c09f7a5c26c78b87b56c67358eb7
+size 6609
diff --git a/config.json b/config.json
new file mode 100644
index 0000000..29e6cbe
--- /dev/null
+++ b/config.json
@@ -0,0 +1,77 @@
+{
+  "architectures": [
+    "Gemma2ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": 50.0,
+  "bos_token_id": 2,
+  "cache_implementation": "hybrid",
+  "dtype": "bfloat16",
+  "eos_token_id": 1,
+  "final_logit_softcapping": 30.0,
+  "head_dim": 256,
+  "hidden_act": "gelu_pytorch_tanh",
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 3584,
+  "initializer_range": 0.02,
+  "intermediate_size": 14336,
+  "layer_types": [
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 8192,
+  "model_type": "gemma2",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 42,
+  "num_key_value_heads": 8,
+  "pad_token_id": 0,
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "sliding_window": 4096,
+  "sliding_window_size": 4096,
+  "transformers_version": "4.56.1",
+  "use_cache": true,
+  "vocab_size": 256000
+}
diff --git a/generation_config.json b/generation_config.json
new file mode 100644
index 0000000..f2c34f2
--- /dev/null
+++ b/generation_config.json
@@ -0,0 +1,9 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 2,
+  "eos_token_id": [
+    1
+  ],
+  "pad_token_id": 0,
+  "transformers_version": "4.56.1"
+}
diff --git a/model-00001-of-00004.safetensors b/model-00001-of-00004.safetensors
new file mode 100644
index 0000000..4705c7e
--- /dev/null
+++ b/model-00001-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0d4eb4fcbddfe01b0dc7a58386a3980826ededaf0ecd67d1ef38995b89615ba
+size 4903351912
diff --git a/model-00002-of-00004.safetensors b/model-00002-of-00004.safetensors
new file mode 100644
index 0000000..744b59b
--- /dev/null
+++ b/model-00002-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712a589c333f49f7b6f3870838a9e0ab4b3c0260abd5130e8c54316e6c99640b
+size 4947570872
diff --git a/model-00003-of-00004.safetensors b/model-00003-of-00004.safetensors
new file mode 100644
index 0000000..48ad5e5
--- /dev/null
+++ b/model-00003-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:87ed143571b20e30b1af91677aa3a0efc1b94fc7a6bc3a66f764ad27ab898d4b
+size 4962221464
diff --git a/model-00004-of-00004.safetensors b/model-00004-of-00004.safetensors
new file mode 100644
index 0000000..f793b72
--- /dev/null
+++ b/model-00004-of-00004.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cced0983c4c03fb39838a94e571ba14bf4058058cdc63d741cc9aa4156d810fe
+size 3670322200
diff --git a/model.safetensors.index.json b/model.safetensors.index.json
new file mode 100644
index 0000000..3b569c5
--- /dev/null
+++ b/model.safetensors.index.json
@@ -0,0 +1,472 @@
+{
+  "metadata": {
+    "total_parameters": 9241705984,
+    "total_size": 18483411968
+  },
+  "weight_map": {
+    "model.embed_tokens.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.10.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.12.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.13.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.14.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.15.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.16.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.17.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.18.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.19.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.20.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.21.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.22.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.24.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.25.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.26.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.27.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.28.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.29.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.3.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.30.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.30.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.input_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.post_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.pre_feedforward_layernorm.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.31.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.32.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.layers.33.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.33.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.34.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.35.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.36.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.37.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.38.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.39.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.4.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.4.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.40.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.40.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.input_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.post_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.pre_feedforward_layernorm.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.41.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
+    "model.layers.5.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.5.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.input_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.post_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.pre_feedforward_layernorm.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.6.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.7.self_attn.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.o_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.7.self_attn.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.post_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.pre_feedforward_layernorm.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.layers.9.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.norm.weight": "model-00004-of-00004.safetensors"
+  }
+}
diff --git a/special_tokens_map.json b/special_tokens_map.json
new file mode 100644
index 0000000..8d6368f
--- /dev/null
+++ b/special_tokens_map.json
@@ -0,0 +1,34 @@
+{
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": {
+    "content": "<bos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<eos>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
diff --git a/tokenizer.json b/tokenizer.json
new file mode 100644
index 0000000..a4a305d
--- /dev/null
+++ b/tokenizer.json
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f7eee611703c5ce5d1eee32d9cdcfe465647b8aff0c1dfb3bed7ad7dbb05060
+size 34362873
diff --git a/tokenizer.model b/tokenizer.model
new file mode 100644
index 0000000..796efe9
--- /dev/null
+++ b/tokenizer.model
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61a7b147390c64585d6c3543dd6fc636906c9af3865a5548f27f31aee1d4c8e2
+size 4241003
diff --git a/tokenizer_config.json b/tokenizer_config.json
new file mode 100644
index 0000000..3ade6be
--- /dev/null
+++ b/tokenizer_config.json
@@ -0,0 +1,2013 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<pad>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "<eos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "<bos>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "<unk>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "<mask>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "5": {
+      "content": "<2mass>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "6": {
+      "content": "[@BOS@]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "7": {
+      "content": "<unused0>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "8": {
+      "content": "<unused1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "9": {
+      "content": "<unused2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "10": {
+      "content": "<unused3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "11": {
+      "content": "<unused4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "12": {
+      "content": "<unused5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "13": {
+      "content": "<unused6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "14": {
+      "content": "<unused7>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "15": {
+      "content": "<unused8>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "16": {
+      "content": "<unused9>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "17": {
+      "content": "<unused10>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "18": {
+      "content": "<unused11>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "19": {
+      "content": "<unused12>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "20": {
+      "content": "<unused13>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "21": {
+      "content": "<unused14>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "22": {
+      "content": "<unused15>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "23": {
+      "content": "<unused16>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "24": {
+      "content": "<unused17>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "25": {
+      "content": "<unused18>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "26": {
+      "content": "<unused19>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "27": {
+      "content": "<unused20>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "28": {
+      "content": "<unused21>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "29": {
+      "content": "<unused22>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "30": {
+      "content": "<unused23>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "31": {
+      "content": "<unused24>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "32": {
+      "content": "<unused25>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "33": {
+      "content": "<unused26>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "34": {
+      "content": "<unused27>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "35": {
+      "content": "<unused28>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "36": {
+      "content": "<unused29>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "37": {
+      "content": "<unused30>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "38": {
+      "content": "<unused31>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "39": {
+      "content": "<unused32>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "40": {
+      "content": "<unused33>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "41": {
+      "content": "<unused34>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "42": {
+      "content": "<unused35>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "43": {
+      "content": "<unused36>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "44": {
+      "content": "<unused37>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "45": {
+      "content": "<unused38>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "46": {
+      "content": "<unused39>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "47": {
+      "content": "<unused40>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "48": {
+      "content": "<unused41>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "49": {
+      "content": "<unused42>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "50": {
+      "content": "<unused43>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "51": {
+      "content": "<unused44>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "52": {
+      "content": "<unused45>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "53": {
+      "content": "<unused46>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "54": {
+      "content": "<unused47>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "55": {
+      "content": "<unused48>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "56": {
+      "content": "<unused49>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57": {
+      "content": "<unused50>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "58": {
+      "content": "<unused51>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "59": {
+      "content": "<unused52>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "60": {
+      "content": "<unused53>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "61": {
+      "content": "<unused54>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "62": {
+      "content": "<unused55>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "63": {
+      "content": "<unused56>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "64": {
+      "content": "<unused57>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "65": {
+      "content": "<unused58>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "66": {
+      "content": "<unused59>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "67": {
+      "content": "<unused60>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "68": {
+      "content": "<unused61>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "69": {
+      "content": "<unused62>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "70": {
+      "content": "<unused63>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "71": {
+      "content": "<unused64>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "72": {
+      "content": "<unused65>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "73": {
+      "content": "<unused66>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "74": {
+      "content": "<unused67>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "75": {
+      "content": "<unused68>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "76": {
+      "content": "<unused69>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "77": {
+      "content": "<unused70>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "78": {
+      "content": "<unused71>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "79": {
+      "content": "<unused72>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "80": {
+      "content": "<unused73>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "81": {
+      "content": "<unused74>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "82": {
+      "content": "<unused75>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "83": {
+      "content": "<unused76>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "84": {
+      "content": "<unused77>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "85": {
+      "content": "<unused78>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "86": {
+      "content": "<unused79>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "87": {
+      "content": "<unused80>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "88": {
+      "content": "<unused81>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "89": {
+      "content": "<unused82>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "90": {
+      "content": "<unused83>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "91": {
+      "content": "<unused84>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "92": {
+      "content": "<unused85>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "93": {
+      "content": "<unused86>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "94": {
+      "content": "<unused87>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "95": {
+      "content": "<unused88>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "96": {
+      "content": "<unused89>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "97": {
+      "content": "<unused90>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "98": {
+      "content": "<unused91>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "99": {
+      "content": "<unused92>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "100": {
+      "content": "<unused93>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "101": {
+      "content": "<unused94>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "102": {
+      "content": "<unused95>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "103": {
+      "content": "<unused96>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "104": {
+      "content": "<unused97>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "105": {
+      "content": "<unused98>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "106": {
+      "content": "<start_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "107": {
+      "content": "<end_of_turn>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "108": {
+      "content": "\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "109": {
+      "content": "\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "110": {
+      "content": "\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "111": {
+      "content": "\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "112": {
+      "content": "\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "113": {
+      "content": "\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "114": {
+      "content": "\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "115": {
+      "content": "\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "116": {
+      "content": "\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "117": {
+      "content": "\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "118": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "119": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "120": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "121": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "122": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "123": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "126": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "127": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "128": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "129": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "130": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "131": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "132": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "133": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "134": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "135": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "136": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "137": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "138": {
+      "content": "\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "139": {
+      "content": "▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "140": {
+      "content": "▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "141": {
+      "content": "▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "142": {
+      "content": "▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "143": {
+      "content": "▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "144": {
+      "content": "▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "145": {
+      "content": "▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "146": {
+      "content": "▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "147": {
+      "content": "▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "148": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "149": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "150": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "152": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "153": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "154": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "155": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "156": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "157": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "158": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "159": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "160": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "161": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "162": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "163": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "164": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "165": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "166": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "167": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "168": {
+      "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "169": {
+      "content": "<table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "170": {
+      "content": "<caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "171": {
+      "content": "<thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "172": {
+      "content": "<tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "173": {
+      "content": "<tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "174": {
+      "content": "<tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "175": {
+      "content": "<th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "176": {
+      "content": "<td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "177": {
+      "content": "</table>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "178": {
+      "content": "</caption>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "179": {
+      "content": "</thead>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "180": {
+      "content": "</tbody>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "181": {
+      "content": "</tfoot>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "182": {
+      "content": "</tr>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "183": {
+      "content": "</th>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "184": {
+      "content": "</td>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "185": {
+      "content": "<h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "186": {
+      "content": "<h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "187": {
+      "content": "<h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "188": {
+      "content": "<h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "189": {
+      "content": "<h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "190": {
+      "content": "<h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "191": {
+      "content": "<blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "192": {
+      "content": "</h1>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "193": {
+      "content": "</h2>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "194": {
+      "content": "</h3>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "195": {
+      "content": "</h4>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "196": {
+      "content": "</h5>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "197": {
+      "content": "</h6>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "198": {
+      "content": "</blockquote>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "199": {
+      "content": "<strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "200": {
+      "content": "<em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "201": {
+      "content": "<b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "202": {
+      "content": "<i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "203": {
+      "content": "<u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "204": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "205": {
+      "content": "<sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "206": {
+      "content": "<sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "207": {
+      "content": "<code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "208": {
+      "content": "</strong>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "209": {
+      "content": "</em>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "210": {
+      "content": "</b>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "211": {
+      "content": "</i>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "212": {
+      "content": "</u>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "213": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "214": {
+      "content": "</sub>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "215": {
+      "content": "</sup>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "216": {
+      "content": "</code>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255968": {
+      "content": "[toxicity=0]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255969": {
+      "content": "\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255970": {
+      "content": "\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255971": {
+      "content": "\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255972": {
+      "content": "\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255973": {
+      "content": "\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255974": {
+      "content": "\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255975": {
+      "content": "\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255976": {
+      "content": "\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255977": {
+      "content": "\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255978": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255979": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255980": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255981": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255982": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255983": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255984": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255985": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255986": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255987": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255988": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255989": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255990": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255991": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255992": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255993": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255994": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255995": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255996": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255997": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255998": {
+      "content": "\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t\t",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "255999": {
+      "content": "<unused99>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<start_of_turn>",
+    "<end_of_turn>"
+  ],
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "extra_special_tokens": {},
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}